06982897 1994982897

라이다, 레이더 및 카메라 센서의 데이터를 사용하는 강화 학습(reinforcement learning)에 근거하는 자율 주행(autonomous travel)시의 최적화된 자원 할당을 위한 방법 및 장치METHOD AND DEVICE FOR OPTIMIZED RESOURCE ALLOCATION DURING AUTONOMOUS TRAVEL ON THE BASIS OF REINFORCEMENT LEARNING WITH USE OF DATA FROM RIDER, RADAR AND CAMERA SENSOR

PROBLEM TO BE SOLVED: To efficiently perform resource allocation during autonomous travel by reinforcement learning and to reduce electric power consumption of an autonomous travel vehicle. SOLUTION: A method for efficiently performing resource allocation during autonomous travel by reinforcement learning includes: a step (a) at which a computing device acquires at least one attention sensor data, and calculates one or more attention scores; a step (b) at which the computing device acquires at least one video data, and generates at least one determination data for autonomous travel; a step (c...

B60W30/095 G06T7/00 G08G1/16 B60W40/04 G06N3/08 B60W60/00

G06K9/00711 G06K9/00791 G06K9/3233 G06K9/6268 G06K9/6277 G06N20/00 G01S13/865 G01S13/867 G01S17/88 G06K9/00791 G06K9/00805 G06N3/006 G06N3/0454 G06N20/00 G01S2013/9318 B60R11/04 B60W30/08 B60W30/14 B60W40/02 B60W2420/42 B60W2420/52 B60Y2300/08 B60Y2300/14 G06F1/329 G06F9/5094 B60R11/04 G01S13/931 G05D1/0221 G05D1/024 G05D1/0248 G05D1/0251 G06N3/08 G06T2207/30261 G06T7/521 G01S17/931

B60W30/095 G06T7/00650B G08G1/16C B60W40/04 G06N3/08 B60W60/00

3D241BA32 3D241BA33 3D241BA42 3D241BA60 3D241BB21 3D241BB28 3D241BB59 3D241BC01 3D241BC02 3D241CC01 3D241CC08 3D241CC17 3D241CD09 3D241CD10 3D241CE05 3D241CE08 3D241CE09 3D241DC01Z 3D241DC02Z 3D241DC03Z 3D241DC06Z 3D241DC21Z 3D241DC25Z 3D241DC26Z 3D241DC27Z 3D241DC31Z 3D241DC33Z 3D241DC43Z 3D241DC45Z 3D241DC50Z 3D241DC58Z 5H181AA01 5H181AA21 5H181AA26 5H181AA27 5H181BB20 5H181CC03 5H181CC04 5H181CC14 5H181FF11 5H181FF27 5H181LL09 5H181LL14 5L096AA06 5L096AA09 5L096BA04 5L096CA05 5L096DA02 5L096FA66 5L096FA69 5L096HA09 5L096HA11 5L096KA04 5L096KA15

32004217 2020004217

32125102 2020125102

2021.12.17

2020.01.15

2020.08.20

2021.11.25

STRADVISION INC

06986290 1994986290

V2X 정보 융합을 이용하여 HD맵을 업데이트하기 위한 자차량 위치 결정(fixing)을 수행하는 방법 및 장치METHOD AND DEVICE FOR MAKING OWN VEHICLE POSITION DETERMINATION TO UPDATE HD MAP BY USING V2X INFORMATION FUSION

PROBLEM TO BE SOLVED: To provide a method capable of calculating the position of an own vehicle so as to enable information acquired by the own vehicle to be more accurately mapped onto an HD map. SOLUTION: In a method for calculating an accurate position of an object vehicle by using relative distance information, when a reference image is acquired through a camera mounted on the object vehicle, a computing device detects a reference object in the reference image S01. The computing device calculates an image base reference distance between the reference object and the object vehicle on the r...

G01C21/30 G08G1/09 G06T7/00 G06T7/70

G01C21/30 G06T2207/30252 G06T7/70 G01C21/32 G01C21/3602 G01S11/12 G01S19/14 G01S19/40 H04W4/40 G05D1/0231 G06N3/08 G06T2207/20081 G06T2207/20084 G06T2210/12 G06T7/50 G06T7/70 G06T7/80 H04W4/40 G06K9/00825 G06K9/3233 G06K9/6217 G06T11/20 G06T2210/12 G06T7/50 H04W4/40

G01C21/30 G08G1/09F G06T7/00350C G06T7/70A

2F129AA03 2F129BB03 2F129BB33 2F129BB49 2F129EE02 2F129EE78 2F129FF02 2F129GG17 2F129HH18 2F129HH19 2F129HH20 5H181AA01 5H181BB04 5H181BB13 5H181BB20 5H181CC04 5H181FF04 5H181FF07 5H181FF13 5H181FF27 5H181LL09 5L096BA04 5L096CA02 5L096DA02 5L096EA35 5L096FA66 5L096FA69 5L096HA09 5L096HA11 5L096JA11

32009529 2020009529

32122785 2020122785

2021.12.22

2020.01.23

2020.08.13

2021.12.01

STRADVISION INC

06980289 1994980289

차선 모델을 이용하여 차선을 검출할 수 있는 학습 방법 및 학습 장치 그리고 이것을 이용한 테스트 방법 및 테스트 장치(test equipment){LEARNING METHOD, LEARNING DEVICE FOR DETECTING LANE USING LANE MODEL AND TEST METHOD, TEST DEVICE USING THE SAME}LEARNING METHOD AND LEARNING DEVICE FOR DETECTING LANE BY USING LANE MODEL, AND TEST METHOD AND TEST DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method that, in detecting a lane from an input image, has less amount of operation than that of a prior art method. SOLUTION: A learning method of a CNN for detecting a lane by using a lane model includes the steps of: a learning device acquiring information on the lanes from an image data set including the lanes; calculating a function parameter of lane modeling functions of the lanes by using the coordinates of the pixels forming the lanes; and classifying the function parameters into K cluster groups by using a clustering algorithm, assigning cluster IDs ...

G06T7/00 G06T7/60

G06K9/00798 G06N3/0454 G06N3/08 G06K9/00798 G06K9/6218 G06N3/0454 G06N3/084 G06T2207/20081 G06T2207/20084 G06T2207/30256 G06T7/73 G06K9/00798 G06N3/084 G06T2207/20081 G06T2207/20084 G06T7/11 G06K9/00798 G06K9/6262 G06K9/6269 G06N3/084 G06T7/77

G06T7/00350C G06T7/60200J

5L096BA04 5L096FA03 5L096FA32 5L096FA62 5L096FA66 5L096GA30 5L096GA32 5L096HA11 5L096JA11 5L096KA04 5L096KA15 5L096MA07

31160073 2019160073

32038661 2020038661

2021.12.15

2019.09.03

2020.03.12

2021.11.19

STRADVISION INC

06986769 1994986769

경로 플래닝을 위한 자율 주행(autonomous travel) 차량의 위치를 결정하는 방법 및 장치METHOD AND DEVICE FOR LOCALIZATION OF AUTONOMOUS VEHICLE FOR ROUTE PLANNING

PROBLEM TO BE SOLVED: To provide a method for detecting a location of a subject vehicle capable of performing an autonomous driving. SOLUTION: A method includes the steps of: regarding one or more live feature maps generated by live image processing corresponding to situations of a subject vehicle, detecting each of feature map coordinates on a live feature map per each of one or more reference objects included in a subject data region corresponding to at least one location and at least one posture of the subject vehicle, by referring to one or more reference feature maps corresponding to the...

G01C21/30 G06T7/00 G06T7/70

G05D1/0214 G05D1/0221 G05D1/0251 G05D1/0276 G05D1/0278 G05D2201/0212 G01C21/20 G01C21/28 G01C21/3602 G06K9/00791 G06K9/6271 G06N3/0454 G06N3/084 G06T2207/20081 G06T2207/20084 G06T2207/30252 G06T7/73 G05D1/0088 G05D1/0268 G05D1/0274 G06N3/08 G06T7/74 G01C21/3602 G06N3/0454 G06N3/084 G06T2207/20081 G06T2207/20084 G06T2207/30252 G06T7/73

G01C21/30 G06T7/00350C G06T7/00650Z G06T7/70Z

2F129AA03 2F129BB03 2F129BB33 2F129BB49 2F129CC14 2F129EE02 2F129EE69 2F129EE78 2F129GG17 2F129HH18 2F129HH19 2F129HH20 5L096BA04 5L096FA09 5L096FA67 5L096FA69 5L096HA11 5L096JA11 5L096KA04 5L096KA15

32008771 2020008771

32122784 2020122784

2021.12.22

2020.01.22

2020.08.13

2021.12.02

STRADVISION INC

06978104 1994978104

GAN를 이용하여 가상 세계에서 가상 데이터로부터 취득한 트레이닝 데이터를 생성해, 자율 주행(autonomous travel)용 뉴럴네트워크의 학습 과정으로 필요한 어노테이션 비용을 삭감하는 학습 방법이나 학습 장치, 그것을 이용한 테스트 방법이나 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR GENERATING TRAINING DATA ACQUIRED FROM VIRTUAL DATA ON VIRTUAL WORLD BY USING GENERATIVE ADVERSARIAL NETWORK (GAN), TO THEREBY REDUCE ANNOTATION COST REQUIRED IN LEARNING PROCESSES OF NEURAL NETWORK FOR AUTONOMOUS DRIVING, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method that transforms virtual images so that virtual images in a virtual world look more real. SOLUTION: A learning method of the present invention includes steps of: (a) applying a convolutional operation to an N-th virtual training image, N-th meta data and (N-K)-th reference information to generate an N-th feature map; (b) applying a deconvolutional operation to the N-th feature map to generate an N-th transformed image; (c) applying a discriminating CNN operation to the N-th transformed image to generate a category score vector; and (d) generat...

G06N3/04 G06N3/08 G06T7/00

G06N3/0454 G06N3/084 G06K9/00791 G06K9/6256 G06N3/0454 G06N3/084 G06N3/10 G05D1/0088 G06K9/00718 G06K9/00791 G06K9/6232 G06K9/6256 G06K9/6277 G06N3/084 G06F16/58

G06N3/04154 G06N3/08 G06T7/00350C

5L096BA04 5L096CA02 5L096CA25 5L096CA27 5L096HA02 5L096HA11 5L096KA04 5L096KA15

32008769 2020008769

32123345 2020123345

2021.12.08

2020.01.22

2020.08.13

2021.11.15

STRADVISION INC

06979228 1994979228

V2V 통신에 의해서 취득된 다른 자율 주행(autonomous travel) 차량의 공간 탐지 결과를 자신의 자율 주행(autonomous travel) 차량의 공간 탐지 결과와 통합하는 학습 방법 및 학습 장치, 그리고 이것을 이용한 테스트 방법 및 테스트 장치(test equipment){LEARNING METHOD AND LEARNING DEVICE FOR INTEGRATING OBJECT DETECTION INFORMATION ACQUIRED THROUGH V2V COMMUNICATION FROM OTHER AUTONOMOUS VEHICLE WITH OBJECT DETECTION INFORMATION GENERATED BY PRESENT AUTONOMOUS VEHICLE, AND TESTING METHOD AND TESTING DEVICE USING THE SAME}LEARNING METHOD AND LEARNING DEVICE FOR INTEGRATING SPACE DETECTION RESULT ACQUIRED THROUGH V2V COMMUNICATION FROM OTHER AUTONOMOUS VEHICLE WITH SPACE DETECTION RESULT GENERATED BY PRESENT AUTONOMOUS VEHICLE, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method for sharing space detection results acquired by using CNN with one another using V2V technology, a testing method, a learning device, and a testing device. SOLUTION: A learning method includes the steps of: a learning device instructing a concatenating network 210 to generate one or more pair feature vectors; the learning device instructing a determining network 220 to apply FC operations to the pair feature vectors, to thereby generate determination vectors and box regression vectors; the learning device instructing a loss unit 230 to genera...

G06N3/08 G06T7/00 G08G1/09 G08G1/16

G06K9/3233 G06K9/6288 G06N3/0454 G06N3/084 G06K9/00805 G06K9/3233 G06K9/6256 G06K9/6262 G06K9/6289 G05D1/0257 G06N3/04 G06N3/084 G06T5/002 G06T7/73 H04W4/46 G06K9/00805 G06K9/6256 G06K9/6267 G06N3/0454 G06N3/08

G06N3/08140 G06T7/00350C G06T7/00650Z G08G1/09H G08G1/16A

5H181AA01 5H181AA21 5H181AA26 5H181AA27 5H181BB04 5H181CC03 5H181CC04 5H181CC14 5L096BA04 5L096FA02 5L096FA69 5L096GA30 5L096GA55 5L096HA11 5L096JA11 5L096KA04

32006088 2020006088

32126623 2020126623

2021.12.08

2020.01.17

2020.08.20

2021.11.17

STRADVISION INC

06979707 1994979707

리그렛션로스(Regression loss)를 이용한 학습 방법 및 학습 장치, 그리고 그것을 이용한 테스트 방법 및 테스트 장치(test equipment){LEARNING METHOD, LEARNING DEVICE USING REGRESSION LOSS AND TESTING METHOD, TESTING DEVICE USING THE SAME}LEARNING METHOD AND LEARNING DEVICE USING REGRESSION LOSS, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method and a learning device using regression loss. SOLUTION: In a method for learning parameters of a CNN on the basis of regression losses, a learning device includes: a step of causing a first to an n-th convolutional layers to generate a first to an n-th encoded feature maps; a step of causing an n-th to a first deconvolutional layers to generate an n-th to a first decoded feature maps from the n-th encoded feature map; a step of generating obstacle segmentation result by referring to a feature of the decoded feature maps; and a step of generati...

G06T7/00 G06N3/08

G06K9/00805 G06K9/4628 G06K9/6273 G06K9/00805 G06K9/4628 G06K9/6273 G06N3/0454 G06N3/084 G06K9/00805 G06N3/084 G06T2207/20081 G06T2207/20084 G06T7/13 G06F17/18 G06K9/34 G06N3/04 G06N3/084 G06N5/046

G06T7/00350C G06N3/08

5L096FA02 5L096FA66 5L096GA30 5L096HA11 5L096KA04

31161679 2019161679

32068015 2020068015

2021.12.15

2019.09.05

2020.04.30

2021.11.18

STRADVISION INC

06979229 1994979229

어안 렌즈가 장착된 2 차원 카메라를 사용해 2 차원 이미지 샷으로 승객 상태를 예측하기 위한 방법 및 장치{METHOD AND DEVICE FOR ESTIMATING PASSENGER STATUSES IN 2 DIMENSION IMAGE SHOT BY USING 2 DIMENSION CAMERA WITH FISHEYE LENS}METHOD AND DEVICE FOR ESTIMATING PASSENGER STATUSES IN 2 DIMENSION IMAGE SHOT BY USING 2 DIMENSION CAMERA WITH FISHEYE LENS

PROBLEM TO BE SOLVED: To provide a method for detecting states of one or more passengers by analyzing a cabin image of a vehicle. SOLUTION: A method includes: a step (a) of, when at least one 2D cabin image of a vehicle cabin photographed by at least one camera with a fisheye lens is acquired, making a passenger state detection device acquire one or more pause points corresponding to one or more passengers from a 2D cabin image with a pause estimation network by inputting the 2D cabin image into the pause estimation network ; and a step (b) of (i) calculating position information related to a...

G08G1/16 G06T7/00

G06K9/00362 G06K9/00838 G06K9/00369 G06K9/00838 G06K9/00845 G06K9/209 G06K2209/40 G06T2207/20084 G06T2207/30196 G06T2207/30268 G06T7/74 G06K9/00369 G06K9/00624 G06K9/46 G06N3/08 G06T3/4007 G06T7/70 H04N7/18 G06K9/00335 G06K9/00838 G06K9/6232 G06T2207/30196 G06T2207/30268 G06T7/74

G08G1/16F G06T7/00660B G06T7/00350C G08G1/16C

5H181AA01 5H181CC04 5H181CC07 5H181LL06 5H181LL20 5L096BA04 5L096FA09 5L096FA66 5L096FA67 5L096FA69 5L096HA11 5L096JA11 5L096KA04

32008409 2020008409

32123342 2020123342

2021.12.08

2020.01.22

2020.08.13

2021.11.17

STRADVISION INC

06970460 1994970460

모바일 디바이스 또는 고정밀도의 소형 네트워크(network)에 적용 가능한 하드웨어(hardware)를 최적화하기 위해서 이용되는 CNN 파라미터(parameter) 및 특징치를 양자화하기 위해서, 무게량자화 로스치를 이용하여 FL(Fractional Length) 치를 결정하는 방법 및 장치{METHOD AND DEVICE FOR DETERMINING FL VALUE BY USING WEIGHTED QUANTIZATION LOSS VALUES TO THEREBY QUANTIZE CNN PARAMETERS AND FEATURE VALUES TO BE USED FOR OPTIMIZING HARDWARE APPLICABLE TO MOBILE DEVICES OR COMPACT NETWORKS WITH HIGH PRECISION}METHOD AND DEVICE FOR DETERMINING FL (FRACTIONAL LENGTH) VALUE BY USING WEIGHTED QUANTIZATION LOSS VALUES TO THEREBY QUANTIZE CNN PARAMETERS AND FEATURE VALUES TO BE USED FOR OPTIMIZING HARDWARE APPLICABLE TO MOBILE DEVICES OR COMPACT NETWORKS WITH HIGH PRECISION

PROBLEM TO BE SOLVED: To optimize hardware applicable to a mobile device or to an accurate and small network. SOLUTION: In the determination of an FL parameter corresponding to the size of an LSB in a quantified value and the size of an FL value, an original vector including an original value is acquired, a quantification operation is applied to the original value with reference to a BW value (BW parameter corresponding to the number of bits used to show each quantified value) and an FL candidate value, each quantified vector (corresponding to FL candidate values and including a quantified va...

G06N3/063 G06F7/499 G06F17/10 G06F17/16

G06N3/0454 G06N3/08 G06N3/0454 G06N3/063 G06N3/04 G06N3/08 G06F17/16 G06N3/04 G06N3/08

G06N3/063 G06F7/499101 G06F17/10A G06F17/16Z

5B056BB26 5B056BB81 5B056BB82

32001009 2020001009

32119528 2020119528

2021.11.24

2020.01.07

2020.08.06

2021.11.02

STRADVISION INC

06970461 1994970461

스마트폰(smart phone), 무선 조정 무인기, 선박 또는 군사적 목적을 위한, 트레이닝 이미지의 최적화 산프링(sampling)에 따라 입력 데이터(input data)를 분석하는 뉴럴네트워크의 온 디바이스 계속 학습 방법 및 장치, 그리고, 이것을 이용한 테스트 방법 및 장치ON-DEVICE CONTINUOUS LEARNING METHOD AND DEVICE OF NEURAL NETWORK FOR ANALYZING INPUT DATA BY OPTIMAL SAMPLING OF TRAINING IMAGE FOR SMART PHONE, DRONE, SHIP, OR MILITARY PURPOSE, AND TEST METHOD AND DEVICE USING IT

PROBLEM TO BE SOLVED: To allow the invasion of privacy to be prevented, resources such as storage to be optimized, and a training image sampling process to be optimized in on-device continuous learning of a neural network that analyzes input data. SOLUTION: A learning device 100 includes a stage 130 that uniform samples new data to have a first volume, causes a boosting network to repeat a process that outputs data prior to synthesis of a second volume corresponding to a k-dimensional modification vector and corresponding to previous data used for learning for a k-dimensional random vector, a...

G06N3/08 G06N3/04

G06K9/6256 G06N3/0454 G06N3/084 G06K9/6257 G06K9/6262 G06N3/0454 G06N3/0472 G06N3/084 G06K9/20 G06N3/04 G06N3/084 G06K9/6257 G06N3/0454 G06N3/084 G06N20/00

G06N3/08 G06N3/04

32007173 2020007173

32123337 2020123337

2021.11.24

2020.01.20

2020.08.13

2021.11.02

STRADVISION INC

06964354 1994964354

최근의 주행 경로에 대한 궤적 기반 행동 분석을 수행해, 차량의 모드를 수동 주행 모드로부터 자율주행모드(autonomous travel)로 변경하는지 아닌지를 판단하는 학습 방법 및 학습 장치LEARNING METHOD AND LEARNING DEVICE FOR DETERMINING WHETHER VEHICLE MODE IS CHANGED FROM MANUAL TRAVEL MODE TO AUTONOMOUS TRAVEL MODE OR NOT BY PERFORMING TRAJECTORY-BASED BEHAVIOR ANALYSIS FOR RECENT TRAVEL ROUTE

PROBLEM TO BE SOLVED: To provide a learning method for calculating a collision probability used in order to determine whether travel mode changeover of a vehicle, which is capable of autonomous travelling, is suitable or not by analyzing a current travel route of an operator. SOLUTION: A method includes: a step (a) at which when state vector and trajectory vector are acquired, a learning device performs (i) a process for generating a state feature map with a state network, and (ii) a process for generating a trajectory feature map with a trajectory network; a step (b) at which the learning de...

B60W30/095 B60W30/14 G08G1/00 G08G1/16 G06T7/00 G06N3/04 G06N3/08

B60W30/095 B60W50/0097 G06N3/0454 G06N3/084 G08G1/0129 G08G1/0133 G08G1/163 G05D1/0061 G06N3/0427 G06N3/0454 G06N3/084 B60W30/0956 B60W60/00 B60W60/001 B60W60/0011 B60W60/0015 B60W60/005 B60W60/0051 B60W60/0053 B60W60/0059 G06K9/00791 G06K9/6232 G06N3/0454 G06N3/08 H04W4/40

B60W30/095 B60W30/14 G08G1/00D G08G1/16A G06T7/00350C G06T7/00650Z G06N3/04154 G06N3/08140

3D241BA01 3D241BA15 3D241BA29 3D241BA49 3D241CD10 3D241CE04 3D241CE05 3D241CE08 3D241CE10 3D241DB01Z 3D241DB02Z 3D241DB20B 3D241DB20Z 3D241DD05B 3D241DD05Z 5H181AA01 5H181BB19 5H181BB20 5H181FF05 5H181FF10 5H181LL01 5H181LL02 5H181LL04 5L096BA04 5L096CA02 5L096CA27 5L096DA02 5L096GA30 5L096HA05 5L096HA11

32006497 2020006497

32121714 2020121714

2021.11.10

2020.01.20

2020.08.13

2021.10.21

STRADVISION INC

06964353 1994964353

하드웨어(hardware) 최적화에 사용되는 이미지 인식을 위한 1 x1 컨벌루션을 이용한 CNN 학습 방법 및 학습 장치, 그것을 이용한 테스트 방법 및 테스트 장치(test equipment){LEARNING METHOD AND LEARNING DEVICE FOR CONVOLUTIONAL NEURAL NETWORK USING 1×1 CONVOLUTION FOR IMAGE RECOGNITION TO BE USED FOR HARDWARE OPTIMIZATION, AND TESTING METHOD AND TESTING DEVICE USING THE SAME}CNN LEARNING METHOD AND LEARNING DEVICE FOR CONVOLUTIONAL NEURAL NETWORK USING 1��1 CONVOLUTION FOR IMAGE RECOGNITION TO BE USED FOR HARDWARE OPTIMIZATION AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method for decreasing an amount of computation in CNN for image recognition, learning device, testing method, and testing device. SOLUTION: The learning method includes the steps of: a learning device causing a first transposing layer 123 or a pooling layer 122 to generate an integrated feature map by concatenating each of pixels, per each of ROIs, in corresponding locations on pooled ROI feature maps; causing a first 1x1 convolution layer 124 and a second 1x1 convolution layer 125 to generate a second adjusted feature map whose volume is adjusted; ...

G06T7/00 G06N3/04 G06N3/08

G06K9/4671 G06K9/6268 G06N3/0454 G06N3/084 G06K9/3233 G06K9/4628 G06K9/6267 G06N3/0454 G06N3/084 G06K9/20 G06N3/0454 G06N3/084 G06K9/6257 G06K9/6262 G06K9/6268 G06N3/04

G06T7/00350C G06N3/04 G06N3/08

5L096EA03 5L096FA02 5L096GA55 5L096HA11 5L096KA04

32002364 2020002364

32119542 2020119542

2021.11.10

2020.01.09

2020.08.06

2021.10.21

STRADVISION INC

06957050 1994957050

모바일 디바이스 또는 소형 네트워크(network)에 적용 가능한 하드웨어(hardware)를 최적화함에 이용 가능한 ROI를 풀링하기 위해서, 마스킹 파라미터를 이용하는 방법 및 장치, 그리고 이것을 이용한 테스트 방법 및 테스트 장치(test equipment){LEARNING METHOD AND LEARNING DEVICE FOR POOLING ROI BY USING MASKING PARAMETERS TO BE USED FOR MOBILE DEVICES OR COMPACT NETWORKS VIA HARDWARE OPTIMIZATION, AND TESTING METHOD AND TESTING DEVICE USING THE SAME}METHOD AND DEVICE USING MASKING PARAMETERS FOR POOLING ROI USABLE FOR OPTIMIZING HARDWARE APPLICABLE TO MOBILE DEVICES OR COMPACT NETWORKS, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for pooling a region of interest (ROI) by using masking parameters. SOLUTION: A method includes, in a computing device, steps of: if an input image is acquired, allowing a convolutional layer of a CNN to generate a feature map corresponding to the input image; allowing an RPN of the CNN to determine an ROI corresponding to at least one object included in the input image by using the feature map; allowing an ROI pooling layer of the CNN to apply each of pooling operations correspondingly to each of sub-regions in the ROI by referring to each of the mas...

G06N3/08 G06N3/04 G06T7/00

G06N3/0454 G06N3/084 G06K9/3233 G06K9/4628 G06K9/6273 G06N3/0454 G06N3/084 G06T2207/20084 G06T7/11 H04N1/64 G06N3/04 G06N3/084 G06T5/20 G06K9/6231 G06K9/6232 G06K9/6256 G06K9/6261 G06K9/6262 G06N3/08

G06N3/08 G06N3/04 G06T7/00350C

5L096FA32 5L096GA10 5L096HA11 5L096JA11 5L096KA04

32001023 2020001023

32119529 2020119529

2021.11.02

2020.01.07

2020.08.06

2021.10.08

STRADVISION INC

06959667 1994959667

자율 주행(autonomous travel) 차의 레벨 4 및 레벨 5를 채우기 위해서 요구되는 ADAS, V2X, 혹은 운전자의 안전을 위해서 이용되는 주행 상황에 대한 정보를 참조해 LDW 알람(alarm)의 발생을 학습하기 위한 방법 및 장치METHOD AND DEVICE FOR LEARNING OCCURRENCE OF LDW ALARM WITH REFERENCE TO INFORMATION ON DRIVING CONDITION USED FOR ADAS, V2X, OR DRIVER SAFETY REQUIRED FOR SATISFYING LEVEL 4 AND LEVEL 5 OF AUTONOMOUS VEHICLE

PROBLEM TO BE SOLVED: To provide a method and a device for use in an autonomous vehicle capable of reducing a driver's accident probability. SOLUTION: In a method for generating an LDW alarm by referring to information on a driving condition, which is used for driver safety, V2X or ADAS required for satisfying a level 4 and a level 5 of an autonomous vehicle, a computing device collects the information on the driving condition including information about whether or not a specific point corresponding to a side mirror on a side of a roadway on which a driver is going to enter belongs to a virtu...

G08G1/167 B60Q9/008 B60R1/00 B60R2300/804 B60W30/18163 B60W50/14 G06K9/00798 G08G1/162 G08G1/167 B60Q1/34 B60R1/06 B60R1/12 B60R2001/1253 B60W30/18163 B60W40/06 B60W40/10 B60W50/14 B60Y2300/18166 B60W2552/00 B60W2556/50 B60Q9/008 B60R1/006 B60R11/04 B60R2300/804 G06K9/00798 G08G1/162 G08G1/167

G08G1/16C

5H181AA01 5H181BB04 5H181BB13 5H181FF04 5H181FF27 5H181LL02 5H181LL04 5H181LL06

32004557 2020004557

32126611 2020126611

2021.11.02

2020.01.15

2020.08.20

2021.10.12

STRADVISION INC

06941386 1994941386

PROBLEM TO BE SOLVED: To provide a method for achieving better performance in an autonomous driving while saving computing powers, by using confidence scores representing a credibility of an object detection which is generated in parallel with an object detection process. SOLUTION: A method includes the steps of: acquiring at least one circumstance image on surroundings of a subject vehicle, through at least one image sensor installed on the subject vehicle (S100); instructing a Convolutional Neural Network (CNN) to apply at least one CNN operation to the circumstance image, to thereby genera...

G06T7/00 G08G1/16

G06K9/00785 G06K9/3233 G06K9/6268 G06N3/0454 G06N3/084 G06K9/00201 G06K9/00805 G06K9/4628 G06N3/0454 G06N3/084 H04W4/40 Y02D30/70 B60W40/02 B60W60/0015 B60W2554/00 G06K9/3233 G06N3/08 G06T2207/20081 G06T2207/20084 G06T7/11 H04W4/40 G06N3/08 H04W4/44

G06T7/00350C G06T7/00650Z G08G1/16A

5H181AA01 5H181AA21 5H181BB04 5H181CC03 5H181CC04 5H181CC12 5H181CC14 5H181LL01 5H181LL04 5H181LL09 5L096BA04 5L096FA02 5L096FA66 5L096GA51 5L096GA55 5L096HA11 5L096JA11 5L096KA04

32010140 2020010140

32126640 2020126640

2021.09.29

2020.01.24

2020.08.20

2021.09.08

STRADVISION INC

06938056 1994938056

경쟁적인 컴퓨팅 및 정보 퓨전을 이용하여, 차량의 자율 주행(autonomous travel)을 지원함에 이용되는 자율 주행(autonomous travel) 서비스 플랫폼(platform)을 제공하는 방법 및 이것을 이용한 서버{METHOD FOR PROVIDING AUTONOMOUS DRIVING SERVICE PLATFORM TO BE USED FOR SUPPORTING AUTONOMOUS DRIVING OF VEHICLES BY USING COMPETITIVE COMPUTING AND INFORMATION FUSION, AND SERVER USING THE SAME}METHOD FOR PROVIDING AUTONOMOUS DRIVING SERVICE PLATFORM TO BE USED FOR SUPPORTING AUTONOMOUS DRIVING OF VEHICLES BY USING COMPETITIVE COMPUTING AND INFORMATION FUSION, AND SERVER USING THE SAME

PROBLEM TO BE SOLVED: To provide an autonomous driving service platform and a method for an autonomous driving vehicle by using a competitive computing and an information fusion. SOLUTION: A method by a service server, includes: a step of acquiring individual sensor data and individual driving data by sensors mounted onto at least a part of autonomous driving vehicles including an object vehicle; a step of carrying out a process for acquiring each autonomous driving source information for the object vehicle and a process for acquiring situation specialization performance information for a dat...

G08G1/00 B60W50/04 G06N3/04 B60W30/00 B60W60/00

G08G1/096725 G08G1/096775 G08G1/0112 G08G1/0141 G08G1/096725 G08G1/096775 B60W40/02 B60W2050/0008 B60W2300/00 B60W2420/00 B60W60/001 B60W2556/35 G05D1/0088 G05D2201/0213 G06N3/08 G05D1/0088 G05D1/0221 G05D2201/0213 G06N3/08

G08G1/00D B60W50/04 G06N3/04 B60W30/00 B60W60/00

3D241BA63 3D241CA00 3D241CD11 3D241CD12 3D241CE02 3D241CE08 3D241DC42Z 3D241DC51Z 3D241DC60Z 5H181AA01 5H181BB04 5H181BB13 5H181BB20 5H181FF10 5H181MC27

32010699 2020010699

32126643 2020126643

2021.09.22

2020.01.27

2020.08.20

2021.09.03

STRADVISION INC

06938049 1994938049

메타 학습을 위해서 잔차 네트워크(network)가 제공되는 CNN의 파라미터(parameter)를 조정하기 위한 학습 방법 및 학습 장치, 그리고 이것을 이용한 실험 방법 및 실험 장치{LEARNING METHOD AND LEARNING DEVICE FOR ADJUSTING PARAMETERS OF CNN IN WHICH RESIDUAL NETWORKS ARE PROVIDED FOR META LEARNING, AND TESTING METHOD AND TESTING DEVICE USING THE SAME}LEARNING METHOD AND LEARNING DEVICE FOR ADJUSTING PARAMETERS OF CNN IN WHICH RESIDUAL NETWORKS ARE PROVIDED FOR META LEARNING, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a CNN-based learning method for meta learning to learn a learning method. SOLUTION: A learning device includes a convolution layer generating an output feature map by applying convolution operation to an image or an input feature map corresponding thereto, and residual networks detouring the convolution layer or a sub-convolution layer corresponding thereto and feed-forwarding an image or an input feature map corresponding thereto to the next convolution layer. The method includes selecting a residual network dropped out from the residual networks, generating ...

G06N3/04 G06N3/08 G06T7/00

G06N3/0454 G06N3/082 G06K9/6262 G06K9/6271 G06N3/0454 G06N3/0481 G06N3/082 G06N3/084 G06N3/04 G06N3/084 G06K9/6232 G06K9/6256 G06K9/6262 G06N3/084

G06N3/04 G06N3/08120 G06T7/00350C

5L096AA02 5L096FA19 5L096HA11 5L096KA04

31212164 2019212164

32119506 2020119506

2021.09.22

2019.11.25

2020.08.06

2021.09.03

STRADVISION INC

06938055 1994938055

운전자 상태에 따라 특화형의 캘리브래이션(calibration)을 위해서 운전자 보조 장치(auxiliary apparatus)를 자동으로 조정하는 방법 및 장치METHOD AND DEVICE FOR ADJUSTING DRIVER ASSISTANCE APPARATUS AUTOMATICALLY FOR PERSONALIZATION AND CALIBRATION ACCORDING TO DRIVER`S STATUS

PROBLEM TO BE SOLVED: To provide a method and a device that allow the position of a mirror of a driver seat to be automatically adjusted according to a physical condition of a driver. SOLUTION: A method for adjusting a position of a driver assistance apparatus according to a driver's status includes the steps of: a position adjusting device 1000, inputting upper and lower body images of a driver, acquired after the driver sits on the driver seat of a vehicle and starts an engine of the vehicle, to a pose estimation network, to acquire body key-points, calculate body part lengths to adjust the...

G06T7/70 G06T7/00 B60R1/02 B60R1/062

B60R16/037 G06K9/00362 G06K9/00597 G06N3/0454 G06K9/00597 G06K9/00832 B60R1/025 B60R16/037 G06K9/00201 G06K9/00268 G06K9/00362 G06K9/00597 G06K9/00845 G06N3/082 G06T9/00 B60R1/04 G06F3/013 G06K9/00248 G06K9/00369 G06K9/00845

G06T7/70Z G06T7/00660B G06T7/00350C B60R1/02 B60R1/062

2H043CB00 2H043CD04 2H043CE02 3D053FF03 3D053GG05 3D053GG06 3D053MM08 3D053MM28 3D053MM52 3D053MM54 5L096AA06 5L096BA04 5L096CA04 5L096DA02 5L096FA67 5L096FA69 5L096HA05 5L096HA08 5L096HA11 5L096JA11 5L096JA18

32010079 2020010079

32126638 2020126638

2021.09.22

2020.01.24

2020.08.20

2021.09.03

STRADVISION INC

06935947 1994935947

스마트폰(smart phone)를 이용하는 보행자를 보호하기 위한 선단 보행자 보조 시스템이 제공하는 방법 및 장치METHOD AND APPARATUS FOR PROVIDING ADVANCED PEDESTRIAN ASSISTANCE SYSTEM FOR PROTECTING PEDESTRIAN USING SMARTPHONE

PROBLEM TO BE SOLVED: To provide a method and an apparatus for providing an advanced pedestrian assistance system for protecting a pedestrian using a smartphone. SOLUTION: The method includes the steps in which: a smartphone acquires first information including position information and speed information of a pedestrian and position information and speed information of the smartphone by a position tracking unit; the smartphone acquires second information including a dangerous state of a dangerous area adjacent to the pedestrian and position information and speed information of a dangerous obje...

G08G1/16 G08G1/005 G08G1/09

G08G1/005 G08G1/166 H04M1/72448 H04M1/72457 G01C21/20 G08G1/005 G08G1/095 G08G1/166 G06Q50/265 G08B21/02 G08B31/00 H04M2201/34 H04M2250/10 H04M2250/12 H04M2250/52 H04N7/18 H04M1/72412 G08G1/005 G08G1/07 G08G1/095 G08G1/166

G08G1/16D G08G1/005 G08G1/09F G08G1/09H G08G1/09D

5H181AA01 5H181AA21 5H181BB04 5H181CC04 5H181FF05 5H181FF22 5H181FF33 5H181LL06 5H181LL15 5H181MB12

32004612 2020004612

32126612 2020126612

2021.09.15

2020.01.15

2020.08.20

2021.08.30

STRADVISION INC

06935939 1994935939

멀티 피팅을 적용 했다고 학습 방법 및 학습 장치 및 그것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE APPLYING MULTI-FEEDING, AND TEST METHOD AND TEST DEVICE UTILIZING THE SAME

PROBLEM TO BE SOLVED: To provide a CNN arithmetic method in which detailed information of an input image does not disappear, even when continuous convolution calculation is applied, in a Convolutional Neural Network (CNN). SOLUTION: The CNN according to the present invention has a configuration of utilizing resized information of at least one input image, by using convolution layers, when sequentially creating a main feature map from a large number of convolution layers. The convolution layer inputs a sub feature map created by referring to the input image, and an integrated feature map that ...

G06K9/4628 G06K9/6232 G06K9/6256 G06N3/0454 G06N3/084 G06N3/082 G06T2207/20081 G06T2207/20084 G06N3/04 G06N3/084 G06N5/046 G06T7/10

31163897 2019163897

32047270 2020047270

2021.09.15

2019.09.09

2020.03.26

2021.08.30

STRADVISION INC

06935948 1994935948

자율 주행(autonomous travel) 차량을 학습, 실험 및 검증을하기 위해서 가상 주행 환경에 대하는 도메인(domain) 적응하셨다고 교통 시나리오를 생성하기 위한 방법 및 장치METHOD AND DEVICE FOR CREATING TRAFFIC SCENARIO ADAPTED TO DOMAIN FOR VIRTUAL TRAVELING ENVIRONMENT FOR LEARNING, TESTING, AND VERIFICATION OF AUTONOMOUS TRAVEL VEHICLE

PROBLEM TO BE SOLVED: To provide a method and a device for creating an elaborate traffic scenario in a virtual travel environment on the basis of actual travel data. SOLUTION: A traffic scenario creation device inputs travel data into a situation analyzing device to extract travel environment information in a state where travel data has been acquired by using detailed traffic data corresponding to sequential traffic data collected from a sensor of a data collection vehicle of prior traffic data and an actual travel environment, which correspond to discrete traffic data extracted from existing...

G09B9/05 G09B9/052 G06K9/00791 G06K9/6271 G06K2009/00738 G06N3/006 G06N3/0445 G06N3/0454 G06N3/08 G06N7/005 G08G1/0112 G08G1/0129 G08G1/0133 G08G1/0141 G08G1/166 G05D1/0088 G05D1/0246 G08G1/0104 G06F30/20 G06F2111/18 G06K9/00711 G06K9/00791 G06K2009/00738 G06N3/0445 G06N3/08 G08G1/0129

G08G1/16A

5H181AA01 5H181BB04 5H181BB13 5H181BB20 5H181CC03 5H181CC04 5H181CC12 5H181CC14 5H181FF04 5H181FF13 5H181FF27 5H181LL01 5H181LL02 5H181LL04 5H181LL09 5H181MC17 5H181MC18 5H181MC19

32009531 2020009531

32123351 2020123351

2021.09.15

2020.01.23

2020.08.13

2021.08.30

STRADVISION INC

06935946 1994935946

분할(segmentation) 스코어 맵 및 클러스터링 맵을 이용 해 자율 주행(autonomous travel) 차의 주행 경로를 계획하기 위한 차선을 검출하는 포스트프로세싱(post-processing) 방법 및 장치POST PROCESSING METHOD AND DEVICE FOR DETECTING LANE IN ORDER TO PLAN TRAVEL ROUTE OF AUTONOMOUS VEHICLE UTILIZING SEGMENTATION SCORE MAP AND CLUSTERING MAP

PROBLEM TO BE SOLVED: To provide a method for post processing for lane detection to be smoothly carried out. SOLUTION: The lane detection method includes: acquiring the segmentation score map and clustering map generated by a CNN (S01); referring to the segmentation score map and determining a lane element including a pixel that forms a lane (S02); referring to the lane element, the segmentation score map and the clustering map and generating seed information (S03); referring to the seed information and generating a base model, and referring to it and generating a lane anchor (S04); referring...

G06T7/00 G08G1/16

G05D1/021 G06K9/00798 G06K9/6218 G06T2207/10016 G06T2207/20081 G06T2207/20084 G06T2207/30256 G08G1/04 G08G1/167 G06T7/73 G06K9/00798 G06T2207/20081 G06T2207/20084 G06T2207/30256 G08G1/167 G06T7/73 G06K9/00798 G06N3/04 G06N3/08 G06T7/11 G06K9/00798 G06K9/6218 G06T2207/10016 G06T2207/20084 G06T2207/30256 G08G1/04 G08G1/167 G06T7/73

G06T7/00350C G08G1/16C G06T7/00650A

5H181AA01 5H181BB04 5H181BB13 5H181BB20 5H181CC04 5H181FF27 5H181LL02 5H181LL04 5H181LL09 5L096BA04 5L096DA02 5L096EA02 5L096FA02 5L096FA66 5L096FA69 5L096GA51 5L096GA55 5L096HA11 5L096KA04 5L096MA07

32004556 2020004556

32126610 2020126610

2021.09.15

2020.01.15

2020.08.20

2021.08.30

STRADVISION INC

06931937 1994931937

사람이 주행 데이터를 트레이닝 데이터로서 이용 해, 강화 학습(reinforcement learning)을 지원함으로써 커스터마이즈형 경로 플래닝을 수행하는 학습 방법 및 학습 장치LEARNING METHOD AND LEARNING APPARATUS FOR CARRYING OUT CUSTOMIZED ROUTE PLANNING BY SUPPORTING REINFORCEMENT LEARNING BY USING PERSON`S TRAVELING DATA AS TRAINING DATA

PROBLEM TO BE SOLVED: To provide a learning method and apparatus for optimizing a route planning for each passenger, and a test method and a testing apparatus using the same. SOLUTION: In a learning apparatus, a method includes: a process S01-1 of referring to information on an actual situation vector and an actual operation, which are included in a travel locus, to generate each first adjustment reward; a process S01-2 of referring to information on the actual situation vector and the actual operation with a common reward module to generate a first common reward; a step S01-2 of referring to...

G08G1/16 G08G1/00 G01C21/34 G06N20/00

G06Q10/047 G01C21/3484 G06N3/006 G06N3/0454 G06N3/084 G06N3/088 G06N5/04 G06N7/005 G06Q10/047 B60W30/10 B60Y2300/10 B60W60/001 G05D1/0088 G05D1/0221 G06N3/04 G06N3/084 G05D1/0088 G05D1/0221 G05D2201/0213 G06K9/6262 G06T15/20 G06T17/05

G08G1/16C G08G1/00D G01C21/34 G06N20/00

2F129AA03 2F129DD13 2F129DD15 2F129DD21 2F129DD39 2F129EE02 2F129EE78 2F129EE94 2F129FF02 5H181AA01 5H181BB13 5H181BB20 5H181FF10 5H181FF22 5H181FF27 5H181LL01 5H181LL02 5H181LL04 5H181LL09

32011163 2020011163

32126646 2020126646

2021.09.08

2020.01.27

2020.08.20

2021.08.19

STRADVISION INC

06932395 1994932395

이미지를 분석하기 위해서, 딥 러닝 네트워크(network)로 이용하기 위한 트레이닝 이미지에 대하는 라벨 링 신뢰도를 자동적이게 평가하기 위한 방법 및 이것을 이용 했다고 신뢰도 평가 장치METHOD FOR AUTOMATICALLY EVALUATING LABELING RELIABILITY OF TRAINING IMAGE FOR USE IN DEEP LEARNING NETWORK TO ANALYZE IMAGE, AND RELIABILITY-EVALUATING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for evaluating labeling reliability of training images to be used for learning a deep learning network. SOLUTION: A method includes the steps of: a reliability-evaluating device instructing a similar-image selection network to select validation image candidates with their own true labels having shooting environments similar to those of acquired original images, which are unlabeled images, and instructing an auto-labeling network to auto-label the validation image candidates with their own true labels and the original images; and (i) evaluating a relia...

G06T7/00 G06N3/04 G06N3/08

G06K9/6267 G06N3/0454 G06N3/084 G06K9/00624 G06K9/4628 G06K9/6254 G06K9/6259 G06K9/6264 G06K9/6274 G06N3/0454 G06N3/084 G06N3/04 G06N3/08 G06F17/18 G06K9/4676 G06K9/6256 G06K9/6267 G06N3/0472 G06N3/084 G06T2207/20132

G06T7/00350C G06N3/04 G06N3/08

5L096GA08 5L096GA51 5L096HA11 5L096KA04

32004704 2020004704

32126613 2020126613

2021.09.08

2020.01.15

2020.08.20

2021.08.20

STRADVISION INC

06921443 1994921443

자율 주행(autonomous travel) 모듈(module) 또는 운전자가 스테아링(steering) 의도를 대상 차량이 스티어링 장치에 보다 정확하게 전달하기 위한 방법 및 장치METHOD AND DEVICE FOR TRANSMITTING AUTONOMOUS TRAVEL MODULE OR STEERING INTENTION OF OPERATOR TO STEERING DEVICE OF OBJECT VEHICLE MORE ACCURATELY

PROBLEM TO BE SOLVED: To provide a method for transmitting an autonomous travel mode or steering intention of an operator to a steering device of an object vehicle more accurately. SOLUTION: A method includes: a step (a) at which if an object intention steering signal, which is inputted by an autonomous travel module is acquired at a present time, a computing device selects a specific reference steering reference value corresponding to the object intention steering signal referring to a reference map with a signal adjustment module; and a step (b) at which the computing device generates an ob...

B62D6/00 G06N20/00

B62D15/025 B62D5/0481 B62D6/002 B62D15/0285 B62D15/021 B62D15/029 G05D1/0088

B62D6/00 G06N20/00130

3D232CC20 3D232DC08 3D232DD01 3D232DD02 3D232EB04 3D232EC34 3D232GG01

32011165 2020011165

32125107 2020125107

2021.08.18

2020.01.27

2020.08.20

2021.07.30

STRADVISION INC

06923960 1994923960

자동 주차 시스템을 제공하기 위해서 결정 지점간이 관계 및 결정 지점에 대하는 리그렛션 결과를 이용 해 주차 스페이스를 검출하는 학습 방법 및 학습 장치, 그리고 이것을 이용 했다고 실험 방법 및 실험 장치LEARNING METHOD AND LEARNING DEVICE FOR DETECTING PARKING SPACES BY USING RELATIONSHIP BETWEEN DECISION POINTS AND REGRESSION RESULTS FOR DECISION POINTS INORDER TO PROVIDE AUTOMATIC PARKING SYSTEM, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for autonomously detecting parking spaces, and an automatic parking system for enabling a driver to easily and simply park. SOLUTION: A learning method for detecting unoccupied parking spaces by using probability distributions on decision points of the unoccupied parking spaces and relational linear segment information on relationships among the decision points is provided. By a learning device, the method includes the steps of: allowing a first CNN to apply a first CNN regression operation to a parking circumstance image, to thereby calculate each of...

G06T7/00 G06N3/08 G06N3/04 G08G1/16

B60W30/06 G06N3/0454 G06N3/08 G06K9/00771 G06K9/00812 G06K9/4628 G06K9/4638 G06K9/52 G06K9/627 G06N3/0454 G06N3/084 G06N7/005 G05D1/0088 G05D2201/0213 G06K9/00812 G06N3/0418 G06N3/084 G08G1/141

G06T7/00650A G06T7/00350C G06N3/08 G06N3/04154 G08G1/16C

5H181AA01 5H181CC04 5H181LL02 5H181LL04 5H181LL09 5H181LL17 5L096BA04 5L096CA02 5L096FA59 5L096GA30 5L096HA11 5L096KA04

32008617 2020008617

32123343 2020123343

2021.08.25

2020.01.22

2020.08.13

2021.08.03

STRADVISION INC

06962604 1994962604

극한 상황으로 폴트 톨러런스 및 변동 로버스트 성이 있는 자동차간 통신 기반의 협업적인 맹점(blind spot) 경보 방법 및 장치COOPERATIVE BLIND SPOT ALARM METHOD AND APPARATUS OF VEHICLE-TO-VEHICLE COMMUNICATION INFRASTRUCTURE WITH FAULT TOLERANCE AND FLUCTUATION ROBUSTNESS UNDER EXTREME SITUATION

PROBLEM TO BE SOLVED: To achieve at least one blind spot alarm by using vehicle-to-vehicle communication regardless of a driving environment. SOLUTION: When a rear camera acquires a rear image of a first vehicle, a first blind spot alarm device transmits the rear image to a blind spot monitor to output first blind spot monitoring information for confirming that each neighboring vehicle is positioned in the rear image. When second blind spot monitoring information for confirming that a second vehicle is positioned at a blind spot is achieved from a second blind spot alarm device of the second ...

G08G1/16 G08G1/09 G06T7/00

G06K9/00791 G06N3/0454 G06N3/084 B60R1/00 B60R2300/50 B60R2300/802 B60R2300/8066 G06K9/00791 G06K9/6273 G06N3/0454 G06N3/084 G08G1/167 H04W4/023 H04W4/025 H04W4/46 B60R1/00 B60R2300/105 B60R2300/802 B60W40/02 B60W50/14 B60W2050/0026 B60W2420/42 B60W2556/50 B60W2556/65 G06N3/08 G06T2207/20081 G06T2210/12 G06T7/11 B60Q9/00 B60R11/04 B60R2300/802 B60R2300/8066 G06K9/00718 G06K9/00825 G06K9/6267 G06N3/084 H04W4/46

G08G1/16D G08G1/09H G06T7/00350C G06T7/00650B

5H181AA01 5H181AA21 5H181AA26 5H181BB04 5H181BB20 5H181CC04 5H181CC12 5H181FF05 5H181LL01 5H181LL02 5H181LL04 5H181LL06 5L096BA04 5L096CA05 5L096DA03 5L096FA66 5L096HA09 5L096HA11 5L096JA11 5L096KA04 5L096KA15

32007175 2020007175

32123339 2020123339

2021.11.05

2020.01.20

2020.08.13

2021.10.18

STRADVISION INC

06962605 1994962605

연속 학습을 이용하여 검출 가능한(detectable) 클래스의 종류를 실시간(real time)으로 업데이트 할 수 있는 물체 검출 시스템을 제공하는 방법 및 이것을 이용한 장치METHOD FOR PROVIDING OBJECT DETECTION SYSTEM WITH WHICH DETECTABLE TYPES OF CLASS CAN BE UPDATED IN REAL TIME UTILIZING CONTINUOUS LEARNING, AND DEVICE UTILIZING THE SAME

PROBLEM TO BE SOLVED: To provide an object detection system with which it is possible to update detectable types of classes in real time. SOLUTION: An object detection system 100 includes: a recognizer 110 for generating a specific query vector after generating a specific feature map when at least one input image is acquired; a similarity determination unit for comparing the specific query vector with a data vector and calculating respective first similarity scores between them, and adding a specific partial image to an unknown image database when a specific first similarity score is less tha...

G06K9/00791 G06K9/6215 G06K9/6256 G06F16/50 G06F16/583 G06K9/00791 G06K9/00993 G06K9/4671 G06K9/6256 G06K9/6271 G06K2209/23 G06N3/0454 G06N3/08 G06F16/55 G06F16/5854 G06K9/00026 G06K9/00067 G06K9/00087 G06N3/082 G06K9/00805 G06K9/2054 G06K9/6232 G06K9/6259 G06K2209/21 G06N3/04 G06N3/08 G06F16/55

G06T7/00350C

5L096EA39 5L096GA51 5L096HA07 5L096HA11 5L096JA03 5L096JA11 5L096KA04

32007176 2020007176

32123340 2020123340

2021.11.05

2020.01.20

2020.08.13

2021.10.18

STRADVISION INC

06919925 1994919925

오토바이 라이더용(laser infrared radar) 스마트 헬멧(head-gear)을 이용 해 오토바이가 맹점(blind spot)를 모니터링(monitoring)하는 방법 및 이것을 이용 했다고 맹점(blind spot) 모니터링(monitoring) 장치METHOD OF MONITORING BLIND SPOT OF MOTORCYCLE USING SMART HELMET FOR MOTORCYCLE RIDER AND BLIND SPOT MONITORING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method capable of making a motorcycle rider exactly recognize peripheral situations in a method of monitoring a blind spot, using a smart helmet for motorcycle rider. SOLUTION: A blind spot monitoring device 100 acquires a video image of a first blind spot corresponding to a smart helmet and confirm a first object positioned at the first blind spot. Furthermore, based on sensor information acquired from a GPS sensor, an acceleration sensor and a geomagnetic sensor mounted on the smart helmet, the blind spot monitoring device determines and refers to a smart ...

G08G1/16 G08G1/09 A42B3/30 H04N7/18

A42B3/0406 A42B3/0433 A42B3/046 A42B3/30 A42B3/0426 B60W50/14 B60W2050/146 B60W2300/36 G01S13/931 G02B27/0172 G02B2027/0123 G02B2027/0138 G06K9/00805 G06K9/4628 G06N3/0454 G06N3/084 G08G1/096791 G08G1/167 G01S2013/9315 A42B3/0426 G02B27/0172 G02B2027/0123 G02B2027/0138

G08G1/16A G08G1/09H A42B3/30 H04N7/18J

3B107EA06 3B107EA07 5C054CA01 5C054CE00 5C054FC12 5C054FF06 5C054HA30 5H181AA01 5H181AA05 5H181BB04 5H181CC04 5H181CC14 5H181FF05 5H181FF25 5H181FF32 5H181LL02 5H181LL04 5H181LL07 5H181LL08

32007737 2020007737

32126629 2020126629

2021.08.18

2020.01.21

2020.08.20

2021.07.28

STRADVISION INC

06919924 1994919924

주행 중으로 운전자로부터의 새로운 지시없이, 대상 차량이 군집 주행할 수 있도록 주행 모드를 변경하는 방법 및 장치{METHOD AND DEVICE FOR SWITCHING DRIVING MODES TO SUPPORT SUBJECT VEHICLE TO PERFORM PLATOON DRIVING WITHOUT ADDITIONAL INSTRUCTIONS FROM DRIVER DURING DRIVING}METHOD AND DEVICE FOR SWITCHING DRIVING MODES TO SUPPORT SUBJECT VEHICLE TO PERFORM PLATOON DRIVING WITHOUT ADDITIONAL INSTRUCTIONS FROM DRIVER DURING DRIVING

PROBLEM TO BE SOLVED: To provide a method of changing a traveling mode so that a subject vehicle travels in a platoon by utilizing platoon traveling information around the subject vehicle without an instruction from a driver. SOLUTION: In the method, when a base station server linked to a subject vehicle that is not included in any of platoon traveling groups obtains a platoon traveling request from the subject vehicle, first platoon traveling information through an Nth platoon traveling information are acquired by referring to real-time platoon traveling information DB; a first platoon trave...

G08G1/00 G08G1/09

B60W30/16 B60W50/00 B60W2050/008 B60W30/165 B60W2556/20 B60W2556/55 G08G1/22 G05D1/0295 G05D1/0297

G08G1/00X G08G1/09F

5H181AA01 5H181BB04 5H181BB13 5H181CC12 5H181MA50

32005648 2020005648

32126620 2020126620

2021.08.18

2020.01.17

2020.08.20

2021.07.28

STRADVISION INC

06921442 1994921442

V2X를 가능한 다양한 응용 프로그램(application program)을 이용 해 자율 주행(autonomous travel) 차량이 현재가 주행 의도를 사람에 대해서 시그널링하는 방법 및 장치METHOD AND DEVICE FOR SIGNALING PRESENT DRIVING INTENTION OF AUTONOMOUS TRAVEL VEHICLE TO HUMAN BY USING VARIOUS V2X-ENABLED APPLICATION

PROBLEM TO BE SOLVED: To provide a method capable of effectively transmitting a signal showing the present driving intention of an operating autonomous travel vehicle to a pedestrian and an occupant. SOLUTION: A driving intention signaling device 100 includes: (a) a step for detecting a pedestrian located ahead of an autonomous travel vehicle by using a periphery video image, and determining whether the pedestrian crosses a road by using a virtual crosswalk; and (b) a step for estimating a crosswalking trajectory corresponding to an expected path of the pedestrian with reference to a moving t...

B60Q1/50 G08G1/16 B60W40/02 B60W30/08 B60Q1/52 B60Q5/00

B60W50/00 B60W2050/0075 H04W4/40 B60Q1/503 B60Q1/525 B60Q5/006 B60R2300/8006 B60R2300/8033 B62D15/0265 G06K9/00335 G06K9/00362 G06K9/00604 G06K9/00805 G06K9/00845 G08G1/166 B60Q1/503 B60Q5/00 B60W2520/06 B60W2520/10 B60W2520/105 B60W60/0017 B60W60/00274 B60W2554/4029 G05D1/0212 G05D1/0231

B60Q1/50Z G08G1/16C B60W40/02 B60W30/08 B60Q1/52 B60Q5/00630B B60Q5/00630Z B60Q5/00620Z B60Q5/00640Z B60Q5/00660B B60Q5/00660C B60Q5/00660F B60Q5/00660Z B60Q5/00610Z

3D241BA31 3D241CA00 3D241CC01 3D241CC08 3D241CC17 3D241CD16 3D241CD28 3D241CE04 3D241CE05 3D241DA52Z 3D241DB01Z 3D241DB05Z 3D241DB32Z 3D241DC31Z 3D241DC32Z 3D241DC41Z 3D241DC59Z 3K339AA22 3K339AA31 3K339BA03 3K339BA10 3K339BA22 3K339BA25 3K339BA26 3K339CA21 3K339CA30 3K339EA01 3K339EA02 3K339EA03 3K339EA04 3K339EA05 3K339EA06 3K339GB01 3K339GB11 3K339GB21 3K339GB26 3K339GB30 3K339JA22 3K339KA01 3K339KA03 3K339KA04 3K339KA06 3K339KA39 3K339MA02 3K339MA07 3K339MB05 3K339MC01 3K339MC03 3K339MC22 3K339MC27 3K339MC36 3K339MC43 3K339MC48 3K339MC49 3K339MC52 3K339MC53 3K339MC74 5H181AA01 5H181AA21 5H181CC04 5H181FF05 5H181FF25 5H181FF32 5H181LL01 5H181LL07 5H181LL08

32007738 2020007738

32125106 2020125106

2021.08.18

2020.01.21

2020.08.20

2021.07.30

STRADVISION INC

06912835 1994912835

자율 주행(autonomous travel) 자동차가 레벨 4를 채우기 위해서 요구하시는 HD맵 업데이트로 이용하시는 적어도 하나가 아다프티브로스 중요구분치 맵을 이용 했다고 아텐션드리분・이미지 분할(segmentation) 학습 방법 및 학습 장치, 그리고 이것을 이용 했다고 실험 방법 및 실험 장치ATTENTION DRIVEN IMAGE SEGMENTATION LEARNING METHOD AND LEARNING DEVICE USING AT LEAST ONE ADAPTIVE LOSS WEIGHTING VALVE MAP WHICH IS USED FOR HD MAP UPDATE REQUIRED FOR SATISFYING LEVEL 4 OF AUTONOMOUS TRAVEL VEHICLE, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide an attention driven image segmentation method using at least one adaptive loss weighting value which is used for updating an HD map required for satisfying a level 4 of an autonomous travel vehicle for more correctly detecting a dim object such as a lane or a road sign visible in the distance. SOLUTION: A learning device including a CNN is configured so that, a learning process comprises: a step for using a soft max layer for generating a soft max score; a step for using a loss weight layer for generating a prediction error value, and applying a loss weighting...

G06N3/0454 G06N3/084 G06T2207/20081 G06T2207/30256 G06T7/11 G06K9/00791 G06K9/34 G06K9/4628 G06K9/627 G06N3/0454 G06N3/084 G06T2207/20081 G06T2207/20084 G06T7/11 G06K9/4642 G06K9/6257 G06K9/6262 G06N3/04 G06N7/005 G06T2207/20081 G06T2207/20084 G06T7/11

G06T7/00350C

5L096BA04 5L096HA11 5L096KA04 5L096KA15

32001765 2020001765

32119533 2020119533

2021.08.04

2020.01.08

2020.08.06

2021.07.13

STRADVISION INC

06917080 1994917080

심층 신경망 워크를 기반과 했다고 이미지 분석을 통하고 사람이 상태 및 주행 환경을 식별 해 자율 주행(autonomous travel)을 보조하는 자율 주행(autonomous travel) 보조 안경AUTONOMOUS TRAVEL AUXILIARY SPECTACLES FOR ASSISTING AUTONOMOUS TRAVEL BY IDENTIFYING HUMAN STATE AND TRAVEL ENVIRONMENT THROUGH IMAGE ANALYSIS BASED ON DEEP NEUTRAL NETWORK

PROBLEM TO BE SOLVED: To provide a method for providing travel safety information through auxiliary spectacles which a driver wears. SOLUTION: This method includes: (a) a step of acquiring a visual angle dependent type travel image corresponding to a driver's visual angle from a camera mounted on auxiliary spectacles, acceleration information and gyroscope information from sensors, inputting the visual angle dependent type travel image into a convolution network to create a feature map, inputting the feature map into each of a detection network, a segmentation network and an identification ne...

G08G1/16 G06T7/00 G09G5/00 G06N3/04 G06N3/08

G02B27/017 G05D1/0214 G05D1/0221 G05D1/0223 G05D1/0253 G05D1/0276 G05D2201/0212 B60K35/00 B60W50/14 B60W2050/146 B60W2520/105 B60W2520/125 B60W2520/14 B60K2370/162 B60K2370/177 B60K2370/178 B60K2370/566 G02B27/017 G02B2027/014 G02B2027/0178 G02B2027/0187 G06K9/00536 G06K9/00671 G06K9/00798 G06K9/00805 G06K9/6267 G06T2207/20081 G06T2207/20084 G06T2207/30252 G06T7/246 B60W50/14 B60W2050/146 G02B27/0172 G02B2027/014 G02B2027/0178 G02B2027/0187 G06K9/00671 G06T7/10 G06T7/74

G08G1/16C G06T7/00650 G06T7/00350C G09G5/00550C G06N3/04 G06N3/08 G06N3/04145 G06N3/04154

5C182AA31 5C182AB15 5C182AB25 5C182BA14 5C182BA46 5H181AA01 5H181AA21 5H181BB20 5H181CC04 5H181CC24 5H181FF25 5H181FF33 5H181FF35 5H181LL01 5H181LL04 5H181LL07 5H181LL08 5L096BA04 5L096DA01 5L096DA03 5L096FA03 5L096FA69 5L096HA11 5L096JA11 5L096KA04

32006187 2020006187

32126625 2020126625

2021.08.11

2020.01.17

2020.08.20

2021.07.21

STRADVISION INC

06924517 1994924517

심층 신경망 워크 기반이 다중 패치 조합(combination)을 이용 해 얼굴을 인식 할, 극도가 상황으로 폴트 톨러런스 및 변동 로버스트 성을 향상 시키는 방법METHOD FOR RECOGNIZING FACE USING MULTIPLE PATCH COMBINATION BASED ON DEEP NEURAL NETWORK AND IMPROVING FAULT TOLERANCE AND FLUCTUATION ROBUSTNESS

PROBLEM TO BE SOLVED: To provide a method for face recognition by using a multiple patch combination based on a deep neural network. SOLUTION: A face-recognizing device 100, if a face image with a 1-st size is acquired, inputs the face image into a feature extraction network, to allow the feature extraction network to generate a feature map by applying at least one convolution operation to the face image with the 1-st size, and to generate multiple features by applying sliding-pooling operation to the feature map, wherein the feature extraction network has been learned to extract at least one...

G06K9/00288 G06K9/4642 G06K9/6268 G06N3/0454 G06N3/084 G06K9/00268 G06K9/00288 G06K9/00744 G06K9/4628 G06K9/623 G06K9/6257 G06K9/6262 G06K9/6272 G06K9/629 G06K9/00228 G06K9/00268 G06K9/4642 G06K9/6232 G06K9/6256 G06K9/6262

G06T7/00660A G06T7/00350C

5L096AA06 5L096CA01 5L096DA02 5L096EA03 5L096EA35 5L096EA39 5L096HA08 5L096HA11 5L096JA11 5L096KA04

32006170 2020006170

32126624 2020126624

2021.08.25

2020.01.17

2020.08.20

2021.08.04

STRADVISION INC

06916552 1994916552

주행 도중으로 발생하는 주행 시나리오를 검출 해, 운전자가 주행 습관을 평가하기 위한 정보가 제공하는 방법 및 장치METHOD AND DEVICE FOR DETECTING TRAVEL SCENARIO GENERATED IN MIDDLE OF TRAVEL AND PROVIDING INFORMATION FOR EVALUATING TRAVEL HAVIT OF DRIVER

PROBLEM TO BE SOLVED: To provide a method for analyzing and evaluating a travel habit of a driver so that the driver can improve the travel habit. SOLUTION: There is acquired a specific listed event vector including information on each specific travel event in a specific order as specific components, a specific predicted travel scenario including the specific travel event is detected by applying an RNN operation to the specific components of the specific listed event vector in a specific order by an RNN, an RNN loss is generated with reference to the specific predicted travel scenario and the...

G06N3/04 G06N3/08 G08G1/00 B60W40/08 B60W40/09

G06N3/0445 G06N3/0454 G07C5/0808 G06Q50/30 G06N3/0445 G06N3/0454 G06N3/084 G06N3/088 G08G1/0133 H04W4/40

G06N3/04145 G06N3/08 G08G1/00D B60W40/08 B60W40/09

5H181AA01 5H181AA21 5H181BB19 5H181BB20 5H181FF10 5H181MC04 5H181MC12 5H181MC27 5L096BA04 5L096CA02 5L096CA27 5L096DA02 5L096HA07 5L096HA11 5L096KA04

32006496 2020006496

32123333 2020123333

2021.08.11

2020.01.20

2020.08.13

2021.07.20

STRADVISION INC

06916549 1994916549

군사적 목적, 무선 조정 무인기 또는 로봇으로 이용하시기 위해서 하나 이상이 이전이 배치를 또한 참조 해 모바일 장치 또는 IOT 장치로 적용 가능한 온라인(online) 배치 정규화, 온 디바이스 학습 및 연속 학습을 수행하는 방법과 장치 및 그것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)METHOD AND DEVICE FOR EXECUTING ONLINE BATCH NORMALIZATION, ON-DEVICE LEARNING, AND CONTINUAL LEARNING APPLICABLE TO MOBILE DEVICES OR IOT DEVICES ADDITIONALLY REFERRING TO ONE OR MORE PREVIOUS BATCHES TO BE USED FOR MILITARY PURPOSE, DRONE, OR ROBOT, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for online batch normalization, on-device learning, and continual learning which are applicable to mobile devices, IoT devices, and the like. SOLUTION: A method includes steps of: via a computing device, instructing a convolutional layer to acquire a k-th batch, and to generate feature maps for the k-th batch by applying convolution operations to input images included in the k-th batch respectively; and via, the computing device instructing a batch normalization layer to calculate adjusted averages and adjusted variations of the feature maps for the k...

G06K9/00624 G06K9/4628 G06K9/6274 G06K9/00624 G06K9/4628 G06K9/6274 G06N3/0454 G06N3/084 G06K9/6221 G06K9/6265 G06K9/66

32001764 2020001764

32119532 2020119532

06916549

2021.08.11

2020.01.08

2020.08.06

2021.07.20

STRADVISION INC

06916548 1994916548

자율 주행(autonomous travel) 자동차가 레벨 4를 채우기 위해서 필요한 HD맵과의 합작을 지원하는 임베딩 로스 및 소프트 막스 로스를 이용 해 적어도 하나가 차선을 가진 이미지를 분할(segmentation)하는 학습 방법 및 학습 장치 및 그것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR SEGMENTING IMAGE HAVING AT LEAST ONE LANE BY USING EMBEDDING LOSS TO SUPPORT COLLABORATION WITH HD MAPS REQUIRED TO SATISFY LEVEL 4 OF AUTONOMOUS VEHICLES AND SOFTMAX LOSS, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method for segmenting an image having one or more lanes to be used for supporting collaboration with HD maps required to satisfy level 4 of autonomous vehicles. SOLUTION: A learning method includes steps of: a learning device causing a CNN module to apply convolution operations to the image, thereby generating a feature map, and apply deconvolution operations thereto, thereby generating segmentation scores of each of pixels on the image; to apply Softmax operations to the segmentation scores, thereby generating Softmax scores; and apply multinomial ...

G06T7/00 G06N3/04 G06N3/08

G06N3/0454 G06N3/084 G06T2207/30236 G06T7/136 G06K9/00798 G06K9/34 G06T2207/20081 G06T2207/20084 G06T2207/30256 G06T7/11 G06T2207/20081 G06T2207/20084 G06T2207/30256 G06T7/143

G06T7/00350C G06T7/00650A G06N3/04 G06N3/08

5L096AA06 5L096BA04 5L096FA32 5L096FA33 5L096FA52 5L096GA51 5L096GA55 5L096HA11 5L096JA11 5L096KA04 5L096MA07

32001763 2020001763

32119531 2020119531

06916548

2021.08.11

2020.01.08

2020.08.06

2021.07.20

STRADVISION INC

06913969 1994913969

강화 학습(reinforcement learning)에 근거하여 협동 주행에서 다중 에이전트 센서 퓨전을 수행하는 방법 및 장치METHOD AND APPARATUS FOR EXECUTING MULTIPLE AGENT SENSOR FUSION BY COOPERATIVE TRAVEL BASED ON REINFORCEMENT LEARNING

PROBLEM TO BE SOLVED: To provide a method and apparatus of learning a sensor fusion network for sensor fusion in an autonomous travel vehicle for executing a cooperative travel. SOLUTION: A learning apparatus 100 includes the steps of: inputting travel images including an autonomous travel vehicle, cooperative travel vehicles of m units and a second virtual vehicle and sensor state information on sensors of n units of the cooperative travel vehicles of m units into a sensor fusion network and creating a sensor fusion probability value which is a probability value of transmitting sensor values...

G08G1/00 G08G1/09 G08G1/16 G06N3/04

G06K9/6269 G06K9/6289 G06N3/0445 G06N3/0454 G06N3/08 G06K9/00791 G06K9/6262 G06K9/6289 G06N3/006 G06N3/0445 G06N3/0454 G06N3/08 G06N5/043 H04W4/46 G05D1/0221 G06K9/6259 G06K9/6289 G06N3/049 H04W4/46

G08G1/00X G08G1/09H G08G1/16A G06N3/04145 G06N3/04

5H181AA01 5H181AA21 5H181AA27 5H181BB04 5H181BB20 5H181CC04 5H181FF13 5H181FF17 5H181FF27 5H181LL01 5H181LL04 5H181LL07 5H181LL09

32007152 2020007152

32123335 2020123335

2021.08.04

2020.01.20

2020.08.13

2021.07.15

STRADVISION INC

06910664 1994910664

자동 주차 시스템 내가 세만틱세그멘테이션을 이용 했다고 주차 영역을 검출하기 위한 방법 및 장치METHOD AND DEVICE FOR DETECTING PARKING AREA USING SEMANTIC SEGMENTATION IN AUTOMATIC PARKING SYSTEM

PROBLEM TO BE SOLVED: To provide a method for learning an automatic parking device of a vehicle for detecting an available parking area. SOLUTION: A method includes the steps of: inputting, if a parking lot image of an area nearby the vehicle is acquired, the parking lot image to output a convolution feature map, outputting a deconvolution feature map by deconvoluting the convolution feature map, and outputting segmentation information by masking the deconvolution feature map; inputting the deconvolution feature map to generate relative coordinates of vertices of a specific available parking ...

G06T7/00 B60W30/06 B60W40/02 G08G1/14 G06N3/08

G06K9/00798 G06K9/3233 G06K9/342 G06K9/6262 G06K9/00812 B60Q5/005 B60Q9/002 B60R11/04 B62D15/0285 G06K9/6263 G06N3/084 G06T2207/20081 G06T2207/20084 G06T2207/30264 G08G1/143 G06T7/11

G06T7/00350C G06T7/00650A B60W30/06 B60W40/02 G08G1/14A G06N3/08140

3D241BA21 3D241BA50 3D241CE05 3D241CE08 5H181AA01 5H181CC04 5H181KK01 5H181KK07 5L096AA06 5L096BA04 5L096EA39 5L096EA43 5L096FA32 5L096FA69 5L096GA10 5L096GA55 5L096HA11 5L096JA11 5L096KA04

32009860 2020009860

32126636 2020126636

06910664

2021.07.28

2020.01.24

2020.08.20

2021.07.09

STRADVISION INC

06916551 1994916551

V2X 통신 및 이미지 처리(image processing)를 이용 했다고 정보 융합에 의해 자율 주행(autonomous travel)이 단기 경로를 플래닝하기 위한 방법 및 장치{METHOD AND DEVICE FOR SHORT-TERM PATH PLANNING OF AUTONOMOUS DRIVING THROUGH INFORMATION FUSION BY USING V2X COMMUNICATION AND IMAGE PROCESSING}METHOD AND DEVICE FOR SHORT-TERM PATH PLANNING OF AUTONOMOUS DRIVING THROUGH INFORMATION FUSION BY USING V2X COMMUNICATION AND IMAGE PROCESSING

PROBLEM TO BE SOLVED: To provide a method for planning autonomous driving in a road situation in which a vehicle capable of V2X communication and a vehicle not capable of V2X communication coexist using V2X communication and image processing. SOLUTION: A method comprises: a step in which a computing device corresponding to a target autonomous vehicle acquires recognition information for surrounding vehicles that includes a first group including a first vehicle capable of V2X communication, and a second group including a second vehicle not capable of V2X communication by a planning module; a s...

G08G1/16 G08G1/09 G06T7/00

G05D1/0221 G05D1/0246 G05D1/0276 G05D1/0287 G06K9/00825 G06K9/6256 G06K9/6288 G06K2209/23 G06K2209/25 G06Q10/047 B60W30/095 B60W30/0956 B60W30/18163 B60W2050/0088 B60W2420/42 B60W2520/10 B60W60/00276 B60W2554/4041 B60W2554/4042 B60W2556/65 G06K9/00785 G06K9/00805 G06K9/3233 G06K9/6271 G06K2209/15 G06K2209/23 G08G1/0112 G08G1/0116 G08G1/0175 G08G1/052 G08G1/163 G08G1/164 G08G1/167 B60R11/04 G06K9/00798 G06N3/0427 G06N3/0454 G06N3/08 G06T2207/20084 G06T2207/30256 G06T7/70 H04W4/44

G08G1/16A G08G1/09F G08G1/09H G06T7/00650B G06T7/00350C

5H181AA01 5H181BB04 5H181BB13 5H181BB20 5H181CC04 5H181CC12 5H181LL04 5H181MC19 5L096AA02 5L096AA06 5L096BA04 5L096CA04 5L096FA66 5L096FA69 5L096GA30 5L096GA51 5L096HA11 5L096KA04

32005636 2020005636

32126619 2020126619

06916551

2021.08.11

2020.01.17

2020.08.20

2021.07.20

STRADVISION INC

06908945 1994908945

기능적 안전성을 위해서 극한 상황으로 폴트 톨러런스 및 변동 로버스트 성을 향상 시키도록 테스트 패턴(test pattern)을 이용 해 CNN가 파라미터(parameter)가 무결성을 검증하기 위한 방법 및 장치METHOD AND DEVICE FOR VERIFYING INTEGRITY OF PARAMETERS OF CNN BY USING TEST PATTERN TO ENHANCE FAULT TOLERANCE AND FLUCTUATION ROBUSTNESS IN EXTREME SITUATIONS FOR FUNCTIONAL SAFETY

PROBLEM TO BE SOLVED: To provide a method for verifying integrity of at least one parameter of a convolutional neural network (CNN) by using at least one test pattern so that an autonomous vehicle can be driven safely from hacker threats. SOLUTION: A method for verifying integrity of parameters of a convolutional neural network (CNN) by using a test pattern added to original input includes, in a computing device, steps of: allowing an adding unit of an integrity verification module 200 to generate extended input by adding a test pattern to original input; allowing the CNN to generate output f...

G06F21/51 G06N3/063 G06F17/10 G06F11/22 G06F21/64

G06K9/00791 G06K9/3233 G06K9/6268 G06N3/0454 G06N3/08 G06K9/00791 G06K9/4628 G06K9/6273 G06N3/0454 G06K9/6202 G06K9/6256 G06K9/6262 G06K9/66 G06N3/08

G06F21/51 G06N3/063 G06F17/10A G06F11/22673F G06F21/64

5B048CC02 5B048DD05 5B056BB26

32004559 2020004559

32119556 2020119556

06908945

2021.07.28

2020.01.15

2020.08.06

2021.07.06

STRADVISION INC

06910081 1994910081

협조 주행을 수행하는 각 차량으로부터 취득하셨다고 각 주행 이미지를 통합하는 방법 및 이것을 이용 했다고 주행 이미지 통합 장치METHOD FOR INTEGRATING DRIVING IMAGES ACQUIRED FROM VEHICLES PERFORMING COOPERATIVE DRIVING AND DRIVING IMAGE INTEGRATING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a device for integrating driving images acquired from vehicles performing cooperative driving. SOLUTION: A driving image integrating device 100 includes the steps of: inputting a main driving image to a main object detector 150, generating a main feature map by applying convolution operation via a main convolutional layer, generating main ROIs via a main region proposal network (RPN), generating main pooled feature maps by applying pooling operation via a main pooling layer, and generating main object detection information for the main objects by applying full...

G06T7/00 G08G1/16 G08G1/09

G06K9/00791 G06K9/3233 G06K9/6262 G06N3/0454 G06N3/084 G06K9/00791 G06K9/6288 G06K9/00791 G06K9/6261 G06K9/6262 G06K9/6289 G06N3/0454 G06N3/084

G06T7/00350C G08G1/16A G08G1/09H G06T7/00650B

5H181AA01 5H181BB04 5H181CC04 5H181LL01 5H181LL02 5H181LL04 5L096BA04 5L096CA05 5L096DA02 5L096EA39 5L096HA02 5L096HA08 5L096HA11 5L096JA11 5L096JA16 5L096KA04 5L096KA15

32009978 2020009978

32126637 2020126637

06910081

2021.07.28

2020.01.24

2020.08.20

2021.07.08

STRADVISION INC

06908939 1994908939

장애물이 하단 라인을 기준으로 ROI를 검출하는 학습 방법 및 학습 장치, 그리고 이것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment){LEARNING METHOD, LEARNING DEVICE FOR DETECTING ROI ON THE BASIS OF BOTTOM LINES OF OBSTACLES AND TESTING METHOD, TESTING DEVICE USING THE SAME}LEARNING METHOD AND LEARNING DEVICE FOR DETECTING ROI ON THE BASIS OF BOTTOM LINES OF OBSTACLES, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for learning parameters of a CNN capable of detecting ROIs of nearest obstacles within an input image. SOLUTION: A learning device includes the steps of: causing a first to an n-th convolutional layers to generate a first to an n-th encoded feature maps from an input image; causing an n-th to a first deconvolutional layers to generate an n-th to a first decoded feature maps from the n-th encoded feature map; if a specific decoded feature map is divided into directions of rows and columns, generating obstacle segmentation result by referring to the n-t...

G06K9/00791 G06K9/00805 G06K9/2054 G06K9/3233 G06K9/3241 G06K9/34 G06K9/62 G06N3/0454 G06N3/084 G06N5/046 G06T2207/20081 G06T2207/20084 G06T7/70 G06N20/00 G06K9/00624 G06K9/00791 G06K9/00805 G06K9/3233 G06K9/34 G06K9/4628 G06K9/6256 G06K9/627 G06N3/0454 G06N3/0481 G06N3/084 G06N3/088 G08G1/166 G06K9/00791 G06K9/00805 G06K9/2054 G06K9/3233 G06K9/3241 G06K9/34 G06K9/62 G06N3/0454 G06N3/084 G06N5/046 G06T2207/20081 G06T2207/20084 G06T7/70 G06N20/00

G06T7/00350C

5L096BA04 5L096FA02 5L096FA06 5L096FA18 5L096GA30 5L096HA11 5L096JA11 5L096KA04

31161673 2019161673

32068013 2020068013

2021.07.28

2019.09.05

2020.04.30

2021.07.06

STRADVISION INC

06908943 1994908943

이미지 콘카테네이션과, 타겟(destination) 물체 예측 네트워크(network)를 이용 해, 다중 카메라 또는 사라운드뷰모니타링으로 이용하시는 물체 검출기를 학습하는 방법 및 학습 장치 및 이것을 이용 했다고 실험 방법 및 실험 장치LEARNING METHOD AND LEARNING DEVICE FOR OBJECT DETECTOR TO BE USED FOR MULTI-CAMERA OR SURROUND VIEW MONITORING BY USING IMAGE CONCATENATION AND TARGET OBJECT PREDICTION NETWORK, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for learning parameters of an object detector based on a CNN adaptable to user's requirements such as a key performance index using image concatenation and a target integration network. SOLUTION: A CNN can be redesigned when scales of objects change as focal length or a resolution changes depending on a key performance index. A method includes, in a learning device 100, steps of: allowing an image-manipulating network to generate n manipulated images; allowing an RPN to generate first to n-th object proposals respectively in the manipulated images, an...

G06T7/00 G06N3/04 G06N3/08

G06N3/0454 G06N3/08 G06T2207/10016 G06T7/292 H04N7/18 G06K9/00805 G06K9/3233 G06K9/4628 G06K9/6256 G06K9/627 G06N3/0454 G06N3/084 H04N7/188 G06K9/4642 G06K9/6257 G06K9/6262 G06N3/04 G06N7/005 H04N7/181

G06T7/00350C G06N3/04 G06N3/08

5L096FA16 5L096HA11 5L096JA11 5L096KA04

32002987 2020002987

32119546 2020119546

2021.07.28

2020.01.10

2020.08.06

2021.07.06

STRADVISION INC

06908944 1994908944

다중 카메라 또는 사라운드뷰모니타링으로 이용하시기 위해서, 타겟(destination) 물체 통합 네트워크(network) 및 타겟(destination) 물체 예측 네트워크(network)를 이용 해, 중요 실적평가 지표와 같은 사용자(user) 요구 사항으로 적용 가능한 CNN 기반이 물체 검출기를 학습하는 방법 및 학습 장치 및 이것을 이용 했다고 실험 방법 및 실험 장치LEARNING METHOD AND LEARNING DEVICE FOR OBJECT DETECTOR BASED ON CNN, ADAPTABLE TO USER`S REQUIREMENTS SUCH AS KEY PERFORMANCE INDEX, BY USING TARGET OBJECT INTEGRATION NETWORK AND TARGET OBJECT PREDICTION NETWORK TO BE USED FOR MULTI-CAMERA OR SURROUND VIEW MONITORING, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for learning parameters of an object detector based on a CNN adaptable to user's requirements such as a key performance index by using a target object integration network and a target region prediction network. SOLUTION: A CNN can be redesigned when scales of objects change as focal length or a resolution changes depending on a key performance index. A method includes, in a learning device 100, steps of: allowing a target region prediction network to search for k-th predicted target regions; allowing an RPN to generate k_1-st to k_n-th object proposal...

G06T7/00 G06N3/04 G06N3/08

G06N3/0454 G06N3/084 G06K9/00805 G06K9/3233 G06K9/6262 G06K9/6269 G06K9/6271 G06N3/0454 G06N3/084 H04N7/18 G06K9/4642 G06K9/6257 G06K9/6262 G06N3/084 G06N7/005

G06T7/00350C G06N3/04 G06N3/08 G06N3/08140

5L096FA16 5L096FA69 5L096GA30 5L096HA11 5L096JA11 5L096KA04 5L096MA03

32002988 2020002988

32119547 2020119547

06908944

2021.07.28

2020.01.10

2020.08.06

2021.07.06

STRADVISION INC

06908946 1994908946

거리 예측을 가능한 레이더를 통하고 취득하시는 정보와 카메라를 통하고 취득하시는 정보를 통합하는 센서 융합을 수행하는 것에 의해, 자율 주행(autonomous travel)을 지원하는 뉴럴네트워크를 향상 시키는 학습 방법 및 학습 장치, 그리고 이것을 사용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR IMPROVING NEURAL NETWORK THAT ASSISTS IN AUTONOMOUS TRAVELING BY CARRYING OUT SENSOR FUSION THAT INTEGRATES INFORMATION ACQUIRED THROUGH RADAR CAPABLE OF DISTANCE PREDICTION AND INFORMATION ACQUIRED THROUGH CAMERA, AND TEST METHOD AND TEST DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for learning a CNN using a camera and a radar together so that the CNN works suitably, even when the object depiction rate of a photographed image that is acquired through a camera is low due to that a photographing condition is unsuitable. SOLUTION: The method includes: the learning device causing convolution computation to be applied to a multichannel integrated image by a convolution layer and generating a feature map (S01); causing output computation to be applied to the feature map by an output layer and generating predictive object information (...

G06T7/00 G08G1/16

G01S13/867 G01S13/931 G06K9/00791 G06N3/0454 G06N3/084 G06K9/00805 G06K9/6273 G06K9/6288 G06K9/6289 G01S7/417 G01S13/867 G01S13/931 G06K9/6228 G06K9/6232 G06K9/6256 G06K9/6261 G06K9/6262 G06K9/629 G06N7/005 G06T2207/10028 G06T2207/10044 G06T2207/20076 G06T2207/20081 G06T2207/20084 G06T2207/30252 G06T7/70 G06N20/00

G06T7/00350C G06T7/00650Z G08G1/16C

5H181AA01 5H181BB19 5H181BB20 5H181CC04 5H181CC14 5H181LL01 5H181LL04 5H181LL09 5L096BA04 5L096CA05 5L096FA66 5L096FA69 5L096HA11 5L096KA04

32007739 2020007739

32126630 2020126630

06908946

2021.07.28

2020.01.21

2020.08.20

2021.07.06

STRADVISION INC

06901802 1994901802

이미지 분석에 근거했다고 로버스트 서베일런스(surveillance)를 위한 보행자 검출기가 학습 방법 및 학습 장치, 그리고 그것을 이용하는 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE OF PEDESTRIAN DETECTOR FOR ROBUST SURVEILLANCE BASED ON IMAGE ANALYSIS, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method and a learning device for a pedestrian detector capable of improving detection rate of a pedestrian, and a testing method and a testing device using the same. SOLUTION: A method for learning a pedestrian detector to be used for robust surveillance on the basis of image analysis includes, in a learning device 100, steps of: generating S1 an image patch x12 by cropping each of regions on a training image 11; allowing an adversarial style transformer 130 to generate S2 a transformed image patch G(x)13 by converting each of pedestrians into trans...

G06K9/00369 G06K9/00805 G06K9/6268 G06N3/0454 G06N3/084 G06K9/00362 G06K9/00805 G06K9/4628 G06K9/6267 G06K9/00362 G06K9/6256 G06N3/08 G06N7/00 G06T2207/20081 G06T2207/30196 G06T7/11

G06T7/00650B G06T7/00350C

5L096BA02 5L096CA02 5L096EA13 5L096EA35 5L096HA11 5L096KA04

32004614 2020004614

32119558 2020119558

2021.07.14

2020.01.15

2020.08.06

2021.06.22

STRADVISION INC

06901803 1994901803

폴트 톨러런스 및 변동 로버스트 성을 위한 복수가 뉴럴네트워크를 사용 해 흔들리는 카메라에 따라 생성하셨다고 비디오로부터 지터링을 없애기 위한 학습 방법 및 학습 장치, 그리고 그것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR REMOVING JITTERING ON VIDEO ACQUIRED THROUGH SHAKING CAMERA BY USING A PLURALITY OF NEURAL NETWORKS FOR FAULT TOLERANCE AND FLUCTUATION ROBUSTNESS, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for detecting jittering in videos generated by a shaken camera to remove the jittering on the videos using neural networks, provided for fault tolerance and fluctuation robustness in extreme situations. SOLUTION: A method includes the steps of: a computing device, generating each of t-th masks corresponding to each of objects in a t-th image; generating each of t-th object motion vectors of each of object pixels, included in the t-th image by applying at least one 2-nd neural network operation to each of the t-th masks, each of t-th cropped images, ea...

G06T7/20 G06T7/00 H04N5/232

G06N3/0454 G06N3/084 G06T5/003 G06T2207/10016 G06T2207/20081 G06T2207/20084 G06K9/3233 G06K9/40 G06K9/4628 G06N3/0454 G06N3/084 G06T5/003 G06T2207/10016 G06T2207/20084 G06T7/246 G06F17/11 G06K9/3233 G06N3/0454 G06N3/084 G06T5/003 G06T2207/10016 G06T2207/20081 G06T2207/20084 G06T7/248

G06T7/20100 G06T7/00350C H04N5/232480 H04N5/232290

5C122DA03 5C122DA14 5C122EA12 5C122EA41 5C122EA54 5C122FH07 5C122FH12 5C122HA35 5C122HA48 5C122HA88 5C122HB10 5L096BA04 5L096CA02 5L096DA01 5L096DA02 5L096EA35 5L096EA39 5L096GA08 5L096GA10 5L096HA04 5L096HA11 5L096KA04 5L096KA15 5L096MA03

32005486 2020005486

32126617 2020126617

2021.07.14

2020.01.16

2020.08.20

2021.06.22

STRADVISION INC

06892157 1994892157

V2X 정보 융합 기술에 의해 취득하셨다, 각 물체에 대하는 깊이 예측 정보 및 각 물체에 대하는 클래스 정보를 이용 해 3 D공간을 재구축함으로써 HD맵을 업데이트하는 학습 방법 및 학습 장치, 그리고 이것을 이용 했다고 실험 방법 및 실험 장치LEARNING METHOD AND LEARNING DEVICE FOR UPDATING HD MAP BY REBUILDING 3D SPACE UTILIZING DEPTH PREDICTION INFORMATION FOR EACH OBJECT AND CLASS INFORMATION FOR EACH OBJECT HAVING BEEN ACQUIRED BY V2X INFORMATION FUSION TECHNOLOGY, TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for rebuilding a 3D space and updating a HD map. SOLUTION: The present invention includes: a step in which a learning device causes a coordinate neural network 130 to generate a local feature map and a global feature vector applying coordinate neural network computation to a coordinate matrix; a step in which the learning device causes a determination neural network 140 to generate a first or an N-th prediction suitability score map applying determination neural network computation to an integrated feature map; and a step in which the learning device ...

G06T7/00 G09B29/00

G01C21/32 G01C21/3638 G06N3/084 G06T17/00 G08G1/096725 G06K9/00201 G06K9/00791 G06N3/0454 G06N3/084 G01C21/3691 G06K9/623 G06K9/627 G06K9/628 G06N3/0454 G06N3/084 G06N20/10 G06N20/20

G06T7/00350C G09B29/00Z

2C032HB11 5L096AA06 5L096AA09 5L096BA04 5L096CA05 5L096EA27 5L096HA11 5L096JA11

32009527 2020009527

32123348 2020123348

2021.06.23

2020.01.23

2020.08.13

2021.05.31

STRADVISION INC

06895694 1994895694

AVM 및 강화 학습(reinforcement learning)을 이용 해 자율 주행(autonomous travel)이 안전성을 달성하기 위한 아텐션드리분아르고리즘을 이용 했다고 리소스 할당을 위한 방법 및 장치{METHOD AND DEVICE FOR ATTENTION-DRIVEN RESOURCE ALLOCATION BY USING AVMAND REINFORCEMENT LEARNING TO THEREBY ACHIEVE SAFETY OF AUTONOMOUS DRIVING}METHOD AND DEVICE FOR ATTENTION-DRIVEN RESOURCE ALLOCATION BY USING AVM AND REINFORCEMENT LEARNING TO THEREBY ACHIEVE SAFETY OF AUTONOMOUS DRIVING

PROBLEM TO BE SOLVED: To provide a method for achieving better performance in an autonomous driving while saving computing powers. SOLUTION: A method includes the steps of: a computing device acquiring at least one circumstance image on surroundings of a subject vehicle, through at least one panorama view sensor installed on the subject vehicle (S100); the computing device applying at least one CNN operation to the circumstance image, to thereby generate initial object information and initial confidence information on the circumstance image (S200); and the computing device generating final ob...

G06K9/00791 G06K9/6268 G06N3/0454 G06N3/084 G06N20/00 G06K9/00805 G06K9/4628 G06F17/18 G06K9/00805 G06K9/6267 G06N3/0454 G06N3/0472 G06N3/084 G06T2207/20081 G06T2207/20084 G06T7/70

G06T7/00350C G06T7/00650Z

5L096BA04 5L096GA30 5L096GA51 5L096HA09 5L096HA11 5L096KA04 5L096KA15

32008883 2020008883

32126633 2020126633

06895694

2021.06.30

2020.01.23

2020.08.20

2021.06.10

STRADVISION INC

06890350 1994890350

V2V 통신을 통해 취득 했다고 정보와 카메라를 통해 취득 했다고 정보를 융합 해 카메라가 피치 캘리브래이션(calibration)을 보다 정교하게 수행하는 것에 의해, 카메라 기반이 로버스트 물체 거리(object distance)가 예측을 제공하는 방법 및 이것을 이용 했다고 장치METHOD FOR PROVIDING ROBUST OBJECT DISTANCE ESTIMATION BASED ON CAMERA BY PERFORMING PITCH CALIBRATION OF CAMERA MORE PRECISELY WITH FUSION OF INFORMATION ACQUIRED THROUGH CAMERA AND INFORMATION ACQUIRED THROUGH V2V COMMUNICATION AND DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for precisely estimating distance even if the quality of input images is not good. SOLUTION: In a method for also using additional information acquired through V2V communication for pitch calibration, a computing device performs a process of instructing an initial pitch calibration module to apply a pitch calculation operation to a reference image, to thereby generate an initial estimated pitch, and a process of instructing an object detection network to apply a neural network operation to the reference image, to thereby generate reference object dete...

G06T7/80 G06T7/00

G08G1/096725 G08G1/165 G08G1/166 H04W4/46 G06T2207/30252 G06T7/80 G06K9/00791 G06K9/036 G06K9/2054 G06K9/6215 G06K9/6232 G06K9/6256 H04N5/23206 H04W4/46 H04N5/232127 H04N5/23218

G06T7/80 G06T7/00350C

5L096FA66 5L096FA69 5L096FA70 5L096HA11 5L096KA04

32006495 2020006495

32126626 2020126626

2021.06.18

2020.01.20

2020.08.20

2021.05.27

STRADVISION INC

06930765 1994930765

V2V 연결에 의해 사람이 상태 인식을 활용 해 검출하셨다고 운전자가 이상 상태(abnormality)가 경고하기 위한 방법 및 장치{METHOD AND DEVICE FOR ALERTING ABNORMAL DRIVER SITUATION DETECTED BY USING HUMANS' STATUS RECOGNITION VIA V2V CONNECTION}METHOD AND DEVICE FOR ALERTING ABNORMAL DRIVER SITUATION DETECTED BY USING HUMANS` STATUS RECOGNITION VIA V2V CONNECTION

PROBLEM TO BE SOLVED: To provide a method for warning by detecting an abnormal situation of a driver of a vehicle based on deep learning. SOLUTION: A method includes the steps of: a driver situation detecting device (a) inputting an interior image of a vehicle to a drowsiness detecting network, to detect a facial part of a specific driver, detect an eye part from the facial part, and detect a blinking state of an eye to determine a drowsiness state, and inputting the interior image to a pose matching network, to detect body key-points of the driver, determine whether the body key-points match...

G06T7/00 G08G1/16 G08G1/09 G06N3/08 B60W40/08 B60W50/14

G06K9/00845 G06K9/00892 G08B21/06 H04W4/33 H04W4/46 B60K28/06 B60W2040/0827 B60W2420/42 B60W2540/223 B60W2540/225 B60W2756/10 G06K9/00845 G08G1/096791 B60W40/08 B60W50/14 B60W2040/0827 B60W2040/0863 B60W2050/143 B60R21/01552 B60W2556/65 G06K9/00248 G06K9/00335 G06K9/00845 G06K9/6232 G06K9/6256 G08G1/163

G06T7/00350C G08G1/16F G08G1/09H G06N3/08 G06T7/00660A B60W40/08 B60W50/14

3D241BA50 3D241BA60 3D241BA70 3D241CD03 3D241CE03 3D241CE05 3D241DC01Z 3D241DC18Z 3D241DC25Z 3D241DC31Z 3D241DC33Z 3D241DC41Z 3D241DC50Z 3D241DD03Z 3D241DD04Z 5H181AA01 5H181AA21 5H181BB04 5H181BB13 5H181BB20 5H181CC04 5H181LL20 5L096BA04 5L096DA03 5L096FA09 5L096FA12 5L096FA67 5L096FA69 5L096HA11 5L096KA04

32010700 2020010700

32126644 2020126644

2021.09.01

2020.01.27

2020.08.20

2021.08.16

STRADVISION INC

06890855 1994890855

AVM를 이용 해 자율 주행(autonomous travel)이 안전성을 달성하기 위한 아텐션드리분아르고리즘을 이용 했다고 리소스 할당을 위한 방법 및 장치{METHOD AND DEVICE FOR ATTENTION-DRIVEN RESOURCE ALLOCATION BY USING AVM TO THEREBY ACHIEVE SAFETY OF AUTONOMOUS DRIVING}METHOD AND DEVICE FOR ATTENTION-DRIVEN RESOURCE ALLOCATION BY USING AVM TO THEREBY ACHIEVE SAFETY OF AUTONOMOUS DRIVING

PROBLEM TO BE SOLVED: To provide a method for achieving better performance in an autonomous driving while saving computing powers, by using confidence scores representing a credibility of an object detection which is generated in parallel with an object detection process. SOLUTION: A method includes the steps of: a computing device acquiring at least one circumstance image on surroundings of a subject vehicle, through at least one panorama view sensor installed on the subject vehicle; the computing device instructing a Convolutional Neural Network (CNN) to apply at least one CNN operation to ...

G06K9/00791 G06N3/0454 G06K9/00805 G06K9/3241 G06K9/6228 G05D1/0238 G06K9/00973 G06K9/2054 G06K9/3233 G06K9/6228 G06N3/08

G06T7/00350C G06T7/00650A

5L096BA04 5L096CA04 5L096FA39 5L096JA03 5L096KA04 5L096KA15

32008882 2020008882

32126632 2020126632

06890855

2021.06.18

2020.01.23

2020.08.20

2021.05.28

STRADVISION INC

06892158 1994892158

온 디바이스 독립형 예측에 근거하여 자율 주행(autonomous travel) 차량이 모드를 바꾸는 것에 의해 자율 주행(autonomous travel)이 안전성을 도모하는 학습 방법 및 학습 장치, 그리고 이것을 이용 했다고 실험 방법 및 실험 장치LEARNING METHOD AND LEARNING DEVICE FOR SWITCHING MODES OF AUTONOMOUS VEHICLE BASED ON ON-DEVICE STANDALONE PREDICTION TO THEREBY ACHIEVE SAFETY OF AUTONOMOUS DRIVING, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method for generating parameters which can represent a degree of credibility of an object detection during a process of the object detection. SOLUTION: A learning method includes steps of: a learning device instructing a convolutional layer to generate a convolutional feature map by applying a convolutional operation to a training image; the learning device instructing an anchor layer to generate an RPN confidence map including RPN confidence scores; the learning device instructing an FC layer to generate CNN confidence scores, to thereby generate a...

G06T7/00 G06N3/04 G06N3/08 B60W40/02 B60W50/14

G05D1/0061 G06K9/00791 G06K9/3233 G06K9/6267 G06N3/0454 G06N3/084 G06K9/00805 G06K9/4628 G05D1/0061 G05D1/0088 G05D2201/0213 G06K9/00791 G06K9/6232 G06K9/6262 G06N3/0472 G06N3/08

G06T7/00350C G06T7/00650Z G06N3/04 G06N3/08 B60W40/02 B60W50/14

3D241BA31 3D241CA00 3D241CD03 3D241CD07 3D241CD09 3D241CD12 3D241CE08 3D241DC01Z 3D241DC21Z 3D241DC25Z 3D241DC31Z 3D241DC33Z 3D241DC41Z 3D241DC51B 3D241DC51Z 5L096AA02 5L096AA06 5L096BA04 5L096EA39 5L096GA40 5L096GA51 5L096GA55 5L096HA11 5L096JA11 5L096KA04 5L096MA07

32009528 2020009528

32123349 2020123349

2021.06.23

2020.01.23

2020.08.13

2021.05.31

STRADVISION INC

06895693 1994895693

레인 마스크(Lane Mask)를 사용 해 후처리없이 입력 이미지로 포함되었다고 하나 이상이 차선을 검출하는 방법 및 장치 및 그것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment){METHOD AND DEVICE FOR LANE DETECTION WITHOUT POST-PROCESSING BY USING LANE MASK, AND TESTING METHOD, AND TESTING DEVICE USING THE SAME}METHOD AND DEVICE FOR LANE DETECTION OF ONE OR MORE LANES INCLUDED IN INPUT IMAGE WITHOUT POST-PROCESSING BY USING LANE MASK, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method and a learning device of a method for detecting lanes by using a lane mask, and a testing method and a testing device using the same. SOLUTION: A method for detecting one or more lanes included in an input image by using a lane mask includes steps of: via a learning device, instructing a second CNN to acquire a final feature map, a segmentation score map, and an embedded feature map; generating a masked feature map by referring to the final feature map and the lane mask; generating a lane feature map by applying at least one or more second co...

G06T7/00 G06N3/04 G06N3/08

G06N3/0454 G06N3/084 G06T2207/20081 G06T2207/20084 G06T2207/30256 G06T7/11 G06K9/00798 G06K9/4628 G06K9/6274 G06N3/0454 G06N3/084 G06K9/00798 G06K9/623 G06K9/6232 G06K9/6257 G06K9/6261 G06K9/6262 G06K9/6267 G06N3/08

G06T7/00350C G06T7/00650A G06N3/04 G06N3/08

5L096AA06 5L096BA04 5L096BA18 5L096CA01 5L096DA02 5L096EA31 5L096EA39 5L096FA32 5L096FA33 5L096FA54 5L096FA62 5L096FA69 5L096GA10 5L096GA51 5L096HA09 5L096HA11 5L096JA11 5L096KA04

32000984 2020000984

32119527 2020119527

2021.06.30

2020.01.07

2020.08.06

2021.06.10

STRADVISION INC

06886202 1994886202

가상 주행이 환경으로 사용하시는 도메인(domain) 적응으로 적용되지 얻는 GAN를 이용 해, 리얼 특징 맵과 동일 또는 유사하는 특성을 가진 버추얼 특징 맵을 생성하는 학습 방법 및 학습 장치, 그리고 그것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR GENERATING VIRTUAL FEATURE MAPS HAVING CHARACTERISTICS BEING SAME AS OR SIMILAR TO REAL FEATURE MAPS BY USING GAN CAPABLE OF BEING APPLIED TO DOMAIN ADAPTATION TO BE USED IN VIRTUAL DRIVING ENVIRONMENT

PROBLEM TO BE SOLVED: To alleviate a problem in which sets of training images in a non-RGB format are produced by transforming sets of training images in an RGB format into those in the non-RGB format through a cycle generative adversarial network (GAN) capable of being applied to domain adaptation. SOLUTION: A learning method for deriving virtual feature maps from virtual images having characteristics which are same as or similar to real feature maps derived from real images, by using GAN including a generating network and a discriminating network includes, in a learning device, steps of: al...

G06T7/00 G05D1/02

G06N3/0454 G06N3/084 G06T11/00 G06K9/0063 G06K9/00791 G06K9/4628 G06N3/0454 G06N3/0472 G06N3/084 G06K9/00791 G06K9/46 G06K9/6256 G06K9/6262 G06N3/0454 G06N3/088 G06N5/046

G06T7/00350C G05D1/02H

5H301GG09 5L096BA04 5L096DA02 5L096HA11 5L096JA03 5L096KA04 5L096KA13 5L096KA15

32004266 2020004266

32119554 2020119554

2021.06.16

2020.01.15

2020.08.06

2021.05.18

STRADVISION INC

06886201 1994886201

도메인(domain) 적응으로 적용되지 얻는 사이클(cycle) GAN를 이용 해 리얼 이미지상이 물체 검출로 사용하시는 리얼 세계에서 리얼 이미지를 가상 세계에서 가상 이미지로 변환하는 런타임(run time) 입력 변환을 위한 학습 방법 및 학습 장치, 그리고 그것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR RUNTIME INPUT TRANSFORMATION FOR TRANSFORMING REAL IMAGE ON REAL WORLD INTO VIRTUAL IMAGE ON VIRTUAL WORLD, TO BE USED FOR OBJECT DETECTION ON REAL IMAGES, BY USING CYCLE GAN CAPABLE OF BEING APPLIED TO DOMAIN ADAPTATION, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method, a learning device, a test method, and a test device which can reduce difference between virtual and reality and reduce annotation costs, and which learns runtime input transformation for transforming real images into virtual images. SOLUTION: A learning method includes steps of: allowing a first transformer to transform a first image to a second image, allowing a first discriminator to generate a 1_1-st result, and allowing a second transformer to transform the second image to a third image, having characteristics which are same as or simila...

G06T7/00 G06N20/00

G06N3/0454 G06N3/08 G06T3/0012 G06K9/0063 G06K9/6271 G06N3/0454 G06N3/0472 G06N3/088 G06T11/00 G06F17/11 G06K9/6257 G06N3/0454 G06T15/10 G06T19/006 G06T2207/20081 G06T7/246

G06T7/00350C G06T7/00650A G06N20/00

5L096AA06 5L096BA04 5L096BA18 5L096CA02 5L096DA01 5L096EA39 5L096FA66 5L096GA51 5L096HA11 5L096JA03 5L096JA11 5L096KA04

32004253 2020004253

32119553 2020119553

2021.06.16

2020.01.15

2020.08.06

2021.05.18

STRADVISION INC

06888849 1994888849

연속 학습을 기반과하는 얼굴 인식을 위한 스마트 데이터베이스(database)를 관리하는 방법 및 장치METHOD AND DEVICE FOR MANAGING SMART DATABASE FOR FACE RECOGNITION BASED ON CONTINUOUS LEARNING

PROBLEM TO BE SOLVED: To effectively manage a smart database in which new face images are added in succession. SOLUTION: The present invention includes a step for counting one or more specific face images that correspond to at least one specific person stored in a smart database in which new face images for face recognition are successively stored and confirming whether the counted first precount value satisfies a first set value; and a step, when the first precount value satisfies the first stet value, inputting each specific face image to a neural aggregation network, generating each qualit...

G06T7/00 G06F16/53

G06K9/00268 G06K9/00288 G06F16/51 G06F16/55 G06F16/583 G06K9/00268 G06K9/00288 G06K9/00926 G06K9/6265 G06K9/6267 G06N3/0454 G06N3/08 G06F16/51 G06K9/00268 G06K9/00926 G06K9/6262 G06N3/08 G06F16/51

G06T7/00660A G06T7/00510F G06F16/53

5L096DA02 5L096EA39 5L096FA52 5L096GA51 5L096HA08 5L096HA11 5L096JA03 5L096JA11 5L096JA25 5L096KA04 5L096KA15

32009530 2020009530

32123350 2020123350

06888849

2021.06.16

2020.01.23

2020.08.13

2021.05.24

STRADVISION INC

06875021 1994875021

유용한 학습 데이터를 취사 선별하기 위한 CNN 기반이 학습 방법 및 학습 장치, 그리고 이것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)CNN-BASED LEARNING METHOD AND LEARNING DEVICE SELECTING USEFUL LEARNING DATA, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To automatically select learning data that is necessary for performance improvement of a learning device. SOLUTION: A learning device 400 creates a first feature map using a first CNN module and creates a second feature map using a second CNN module. The learning device further includes: the step of creating a first output indicating identification information and positional information of an object using the first feature map and calculating a first loss by referring to the first output and an original text correct answer image corresponding to the first output; the ste...

G06N3/04 G06T7/00 G06N3/08

G06K9/6256 G06N3/0454 G06N3/084 G06K9/3241 G06K9/4628 G06K9/6255 G06K9/6256 G06N3/0454 G06N3/0481 G06N3/084 G06T2207/20081 G06T2207/20084 G06T7/11 G06K9/38 G06K9/6256 G06K9/78 G06N3/084 G06N5/046

G06N3/04 G06T7/00350C G06N3/08140

5L096FA69 5L096GA30 5L096GA51 5L096HA09 5L096HA11 5L096JA11 5L096KA04

31185244 2019185244

32068028 2020068028

2021.05.19

2019.10.08

2020.04.30

2021.04.26

STRADVISION INC

06872264 1994872264

하드웨어(hardware) 최적화를 통하고 모바일 장치 또는 고정밀도가 소형 네트워크(network)로 사용하시는 CNN 파라미터(parameter) 양자화를 최적화하기 위해서 CNN 레이어를 변환하는 방법 및 장치METHOD AND DEVICE FOR TRANSFORMING CNN LAYERS TO OPTIMIZE CNN PARAMETER QUANTIZATION TO BE USED FOR MOBILE DEVICES OR COMPACT NETWORKS WITH HIGH PRECISION VIA HARDWARE OPTIMIZATION

PROBLEM TO BE SOLVED: To provide a method for transforming CNN layers to flatten values included in at least one feature map in order to properly reflect respective values of specific channels including small values on output values. SOLUTION: A method for transforming convolutional layers of a CNN including m convolutional blocks includes steps of: generating k-th quantization loss values by referring to k-th initial weight values of a k-th initial convolutional layer included in a k-th convolutional block, a (k-1)-th feature map outputted from the (k-1)-th convolutional block, and each of k...

G06T7/00 G06N3/04 G06N3/08

G06F9/30025 G06N3/0454 G06N3/0454 G06N3/063 G06N3/08 G06T3/4046 G06T3/4084

G06T7/00350C G06N3/04 G06N3/08

5L096EA39 5L096HA11 5L096JA11 5L096KA04

31238383 2019238383

32119518 2020119518

2021.05.19

2019.12.27

2020.08.06

2021.04.21

STRADVISION INC

06872260 1994872260

자동차가 차선 변경에 대하는 위험을 경보하는 방법 및 이것을 이용 했다고 경보 장치(warning system)METHOD FOR WARNING DANGER TO LANE CHANGE OF AUTOMOBILE AND WARNING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide algorithm with less quantity of data to be calculated for detecting an automobile located at a blind spot. SOLUTION: A method for warning danger to lane change of an automobile comprises steps in which: a warning device acquires, when a rear image photographed by a traveling automobile is acquired, a segmentation image which is associated with the rear image by performing segmentation of a rear video image by using a learned CNN; and the warning device confirms a free space ratio at a blind spot by referring to a segmentation image, however, in a step in which...

G08G1/16 H04N7/18 G06T7/62 G06T7/00

B60Q9/008 B60R1/00 B60R2300/802 B60W50/14 B60W2050/143 B60Q9/008 G06K9/00805 G06K9/4628 G06K9/6271 G06N3/0454 G06N3/08 G06T2207/20084 G06T2207/30256 G08G1/167 G06T7/11 B60W30/18163 B60W40/02 B60W50/14 B60Y2300/18166 B60Y2400/90 G06T2207/20081 G06T2207/20084 G08G1/167 G06T7/11 B60Q9/008 G06N3/08 G06T7/20 G06T2207/20084 G06T2207/30256 G06T2207/30261 G06T7/11 G06T7/70

G08G1/16C H04N7/18J G06T7/62 G06T7/00650A

5C054CC02 5C054FF06 5C054HA30 5H181AA01 5H181CC04 5H181LL02 5H181LL04 5H181LL06 5H181LL09 5L096BA04 5L096CA04 5L096DA03 5L096FA02 5L096FA59 5L096GA17 5L096GA51 5L096HA02 5L096HA11 5L096KA04 5L096MA07

31183127 2019183127

32077393 2020077393

2021.05.19

2019.10.03

2020.05.21

2021.04.21

STRADVISION INC

06869559 1994869559

장애물을 검출하는 학습 방법 및 학습 장치 그리고 이것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment){LEARNING METHOD, LEARNING DEVICE FOR DETECTING OBSTACLES AND TESTING METHOD, TESTING DEVICE USING THE SAME}LEARNING METHOD AND LEARNING DEVICE FOR DETECTING OBSTACLE, AND TEST METHOD AND TEST DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a parameter learning method for a CNN for detecting an obstacle from a training image. SOLUTION: A parameter learning method includes steps of: acquiring a training image and generating an encoded feature map with a convolution layer; generating a decoded feature map with a deconvolution layer; partitioning the decoded feature map with respect to a direction of the rows and the columns, and when each cell of a grid with rows and columns is generated, concatenating features of the rows per column in a channel direction to generate a corrected feature map; and c...

G06K9/00805 G06K9/6256 G06N3/0454 G06N3/084 G06K9/00805 G06K9/6232 G06N3/0454 G06N3/0481 G06N3/084 G06T2207/20081 G06T2207/20084 G06T2207/30261 G06T7/143 G06K9/00805 G06K9/6262 G06N3/084 G06T2207/20021 G06T7/11 G06T7/70

G06T7/00350C

5L096AA06 5L096BA04 5L096CA04 5L096DA02 5L096HA09 5L096HA11 5L096JA11 5L096JA18 5L096KA04

31160143 2019160143

32038663 2020038663

06869559

2021.05.12

2019.09.03

2020.03.12

2021.04.16

STRADVISION INC

06872259 1994872259

R-CNN 기반이 물체 검출기가 학습 방법과 테스트 방법 및 그것을 이용 했다고 학습 장치와 테스트 장치(test equipment)LEARNING METHOD AND TESTING METHOD FOR R-CNN BASED OBJECT DETECTOR, AND LEARNING DEVICE AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method and a testing method for an object detector characterized in using one R-CNN network that performs learning by using data sets different from each other, and a learning device and a testing device using the same. SOLUTION: A testing device 200 for an object detector includes a step of convolutional layers generating feature maps, an RPN outputting ROI regression information and matching information, a proposal layer outputting ROI candidates, and a proposal selecting layer outputting the ROI proposals; a step of pooling layers calculating fea...

G06N3/04 G06T7/00 G06N3/08

G06K9/3233 G06K9/6201 G06K9/6256 G06K9/6267 G06K9/628 G06N3/0454 G06N3/084 G06K9/3233 G06K9/6262 G06K9/6267 G06K9/6271 G06N3/0454 G06N3/084 G06K9/3241 G06K9/6201 G06K9/6232 G06K9/6257 G06K9/6259 G06K9/6262 G06K9/628 G06N3/08

G06N3/04 G06T7/00350C G06N3/08

5L096BA04 5L096HA11 5L096JA11 5L096KA04

31182920 2019182920

32061141 2020061141

2021.05.19

2019.10.03

2020.04.16

2021.04.21

STRADVISION INC

06903352 1994903352

비최대치 억제를 학습하는 병합 네트워크(network)를 이용 했다고 이종 센서 융합을 위한 학습 방법 및 학습 장치{LEARNING METHOD AND LEARNING DEVICE FOR HETEROGENEOUS SENSOR FUSION BY USING MERGING NETWORK WHICH LEARNS NON-MAXIMUM SUPPRESSION}LEARNING METHOD AND LEARNING DEVICE FOR HETEROGENEOUS SENSOR FUSION BY USING MERGING NETWORK WHICH LEARNS NON-MAXIMUM SUPPRESSION

PROBLEM TO BE SOLVED: To provide a learning method for generating integrated object detection information of an integrated image by integrating first object detection information and second object detection information. SOLUTION: A method includes the steps of: (a) a learning device, if first object detection information and second object detection information is acquired, instructing a concatenating network included in a DNN to generate pair feature vectors including information on pairs of first original ROIs and second original ROIs; (b) the learning device instructing a determining networ...

G06K9/00791 G06K9/3233 G06N3/0454 G06N3/084 G06K9/00791 G06K9/3233 G06K9/6271 G06K2009/2045 G06N3/0454 G06N3/08 G06F17/18 G06K9/00805 G06K9/3233 G06K9/6212 G06K9/6256 G06N3/04 G06N3/0635

G06T7/00350C

5L096BA04 5L096CA05 5L096FA18 5L096FA64 5L096FA69 5L096GA22 5L096GA30 5L096GA51 5L096HA11 5L096JA11 5L096KA04 5L096KA15

32006051 2020006051

32126622 2020126622

2021.07.14

2020.01.17

2020.08.20

2021.06.25

STRADVISION INC

06870868 1994870868

Non-RGB 포맷(format)을 가진 이미지가 물체에 대하는 물체 검출 학습으로 사용하시지 얻도록, 사이클(cycle) GAN를 사용 해 RGB 트레이닝 이미지 셋을 Non-RGB 트레이닝 이미지 셋으로 변환해, 컴퓨팅 부하 및 리사이클(recycling) 데이터를 현저히 감소시킬 수 있는 학습 방법 및 학습 장치, 그리고 그것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE WITH WHICH IT IS POSSIBLE TO CONVERT RGB TRAINING IMAGE SET INTO NON-RGB TRAINING IMAGE SET USING CYCLE GAN SO AS TO BE USABLE FOR OBJECT DETECTION LEARNING FOR OBJECT OF IMAGE HAVING NON-RGB FORMAT AND MARKEDLY REDUCE COMPUTING LOAD AND RECYCLE DATA, AND TEST METHOD AND DEVICE UTILIZING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method for converting an RGB training image set to a Non-RGB training image set so as to be usable for object detection learning of an image having a Non-RGB format. SOLUTION: The learning method includes the steps of: causing a first image having an RBG format to be converted into a second image having a Non-RGB format by a first transformer 130 and causing the second image to be converted into a third image having an RGB format by a second transformer 150; causing a fourth image having a Non-RGB format to be converted into a fifth image having an ...

G06N3/0454 G06N3/08 G06T3/0012 G06K9/36 G06N3/0454 G06N3/0472 G06N3/08 G06K9/4652 G06K9/6256 G06N3/0454 G06T11/001 G06T2207/10024 G06T2207/20081 G06T2207/20084 G06T7/90

G06T7/00350C

5L096AA02 5L096EA45 5L096HA11 5L096KA04

32004204 2020004204

32126606 2020126606

2021.05.12

2020.01.15

2020.08.20

2021.04.19

STRADVISION INC

06869562 1994869562

트래킹(tracking) 네트워크(network)를 포함하는 CNN를 사용 해 물체를 트래킹(tracking)하는 방법 및 그것을 이용 했다고 장치{METHOD FOR TRACKING OBJECT BY USING CONVOLUTIONAL NEURAL NETWORK INCLUDING TRACKING NETWORK AND COMPUTING DEVICE USING THE SAME}METHOD FOR TRACKING OBJECT BY USING CNN INCLUDING TRACKING NETWORK, AND DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for tracking an object by using a CNN. SOLUTION: A method comprises steps of: on a device, acquiring a feature map from a current video frame of a video, and instructing an RPN to generate information on a proposal box (PB) for an object; generating, from a previous state vector of a previous bounding box (BB) for the object located on a previous video frame, an estimated state vector of the previous BB by using a Kalman filter algorithm (KFA); generating an estimated BB in the current video frame tracked from the position of the previous BB for the e...

G06T7/00 G06T7/277

G06N3/0454 G06N3/084 G06T7/277 G06T2207/10016 G06T2207/20081 G06T2207/20084 G06K9/00711 G06K9/3233 G06K9/4628 G06T7/246 G06T7/277 G06T2207/10016 G06T2207/20081 G06T2207/20084 G06K9/00718 G06K9/6256 G06K9/6262 G06T7/248 G06T7/277 G06T2207/10016 G06T2207/20081 G06T2207/20084

G06T7/00350C G06T7/277

5L096AA06 5L096BA04 5L096BA18 5L096CA04 5L096DA01 5L096FA05 5L096FA33 5L096FA59 5L096FA69 5L096HA05 5L096HA11 5L096JA11 5L096KA04

31183394 2019183394

32061144 2020061144

2021.05.12

2019.10.04

2020.04.16

2021.04.16

STRADVISION INC

06868298 1994868298

R-CNN 기반이 물체 검출기를 학습하는 방법 및 테스트하는 방법, 그것을 이용 했다고 학습 장치 및 테스트 장치(test equipment)METHOD FOR LEARNING AND METHOD FOR TESTING R-CNN BASED OBJECT DETECTOR, AND LEARNING DEVICE AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for learning and a method for testing an R-CNN based object detector, and a learning device and a testing device using the same. SOLUTION: A learning device 100 instructs a k-th up-sampling block to a first up-sampling block to output a (k-1)-st pyramidic feature map to a first pyramidic feature map respectively, instructs an RPN to output each ROI corresponding to each candidate region, instructs a pooling layer to output a feature vector; and learns parameters of an FC layer, the k-th up-sampling block to the first up-sampling block, and the first f...

G06T7/00 G06N3/04 G06N3/08

G06K9/46 G06K9/6269 G06N3/0445 G06N3/0454 G06N3/08 G06K9/3233 G06K9/3241 G06K9/4628 G06N3/0454 G06N3/084 G06K9/3233 G06N3/08 G06T2207/20081 G06N20/00

G06T7/00350C G06N3/04 G06N3/08140

5L096AA06 5L096BA08 5L096DA02 5L096HA09 5L096HA11 5L096KA04 5L096KA15

31183128 2019183128

32061143 2020061143

2021.05.12

2019.10.03

2020.04.16

2021.04.14

STRADVISION INC

06869565 1994869565

위험 요소 검출로 이용하시는 학습용무 이미지 데이터(image data) 세트가 생성 방법 및 컴퓨팅 장치, 그리고 이것을 이용 했다고 학습 방법 및 학습 장치{METHOD AND COMPUTING DEVICE FOR GENERATING IMAGE DATA SET TO BE USED FOR HAZARD DETECTION AND LEARNING METHOD AND LEARNING DEVICE USING THE SAME}METHOD AND COMPUTING DEVICE FOR GENERATING IMAGE DATA SET FOR LEARNING TO BE USED FOR HAZARD DETECTION AND LEARNING METHOD AND LEARNING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for generating an image data set for learning to be used in CNN(Convolutional Neural Network) for object detection from an input image. SOLUTION: A method comprises: a step of acquiring a first label image where an edge portion is set in the boundary between an object and the background, and a label value different from each other is assigned corresponding to each of the background and the object; a step of extracting the edge portion, and generating an edge image from the first label image; a step of merging an edge enhancement image generated by imp...

G06T7/00 G06T7/12 G06T7/194 G06N3/08 G06N3/04

G06T5/001 G06T2207/10004 G06T2207/20081 G06T2207/20084 G06T2207/20192 G06T7/13 G06T7/181 G06K9/00624 G06K9/3233 G06K9/34 G06K9/4604 G06K9/4628 G06K9/6256 G06K9/627 G06N3/0454 G06N3/084 G06T2207/20084 G06T2207/30261 G06T7/11 G06T7/194 G05D1/0221 G05D1/0246 G05D2201/0213 G06K9/00805 G06K9/00818 G06K9/6256 G06N3/08 G06N5/046 G06T2207/30261 G06T7/194 G06T7/70 G06N20/00

G06T7/00350C G06T7/12 G06T7/194 G06N3/08 G06N3/04

5L096AA06 5L096BA04 5L096BA18 5L096CA01 5L096DA01 5L096EA04 5L096EA39 5L096FA06 5L096HA11 5L096KA04

31193145 2019193145

32119496 2020119496

06869565

2021.05.12

2019.10.24

2020.08.06

2021.04.16

STRADVISION INC

06864891 1994864891

자율 주행(autonomous travel) 차량 레벨 4를 채우기 위해서 영역이 클래스에 따라 모드를 바꾸기 위해서 그리드(grid) 생성기를 이용하는 뉴럴네트워크 연산 방법 및 이것을 이용 했다고 장치NEURAL NETWORK OPERATION METHOD USING GRID GENERATOR FOR CONVERTING MODES ACCORDING TO CLASSES OF AREAS TO SATISFY LEVEL 4 OF AUTONOMOUS VEHICLES, AND DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a neural network operation method using a grid generator for converting modes according to classes of areas to satisfy level 4 of autonomous vehicles. SOLUTION: A neural network operation method includes steps of: a computing device (a) when a test image is acquired, causing a non-object detector to acquire non-object location information for testing and non-object class information for testing by detecting the non-objects for testing on the test image; (b) causing the grid generator to generate section information by referring to the non-object location infor...

G06T7/00 G06N3/08

G06N3/08 G06N3/10 G06K9/00791 G06K9/627 G06N3/0454 G06N3/08 G06T2207/20021 G06T2207/20081 G06T2207/20084 G06T2207/30252 G06T7/11 G06K9/00791 G06K9/6257 G06K9/6261 G06K9/6262 G06N3/08 G06T2207/20021 G06T2207/20081 G06T2207/20084 G06T2207/30252 G06T7/70

G06T7/00350C G06N3/08 G06T7/00650Z

5L096AA06 5L096BA04 5L096EA39 5L096FA59 5L096FA69 5L096GA51 5L096GA55 5L096HA11 5L096JA09 5L096JA11 5L096KA04 5L096MA07

32002045 2020002045

32119538 2020119538

06864891

2021.04.28

2020.01.09

2020.08.06

2021.04.07

STRADVISION INC

06865342 1994865342

CNN 기반 차선 검출을 위한 학습 방법 및 학습 장치, 그리고 이것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR DETECTING LANE ON THE BASIS OF CNN, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a leaning method and a learning device for detecting a lane accurately on the basis of CNN, and a testing method and a testing device. SOLUTION: A learning method includes: applying convolution operations to an image, to generate lane candidate information; and generating a first pixel data map including information on pixels in the image and their corresponding pieces of first data. Main subsets selected from the first data include values generated by calculating distance from the pixels to their nearest first lane candidates. The method includes generating a...

G06K9/00798 G06K9/342 G06N3/0445 G06N3/0454 G06N3/08 G06K9/00798 G06K9/6262 G06N3/0454 G06N3/084 G06K9/00798 G06K9/6262 G06N3/08

G06T7/00350C G06T7/00650A

5L096AA06 5L096BA04 5L096CA01 5L096DA01 5L096FA66 5L096FA69 5L096GA30 5L096GA51 5L096HA11 5L096JA11 5L096KA04

31163899 2019163899

32047272 2020047272

2021.04.28

2019.09.09

2020.03.26

2021.04.08

STRADVISION INC

06867054 1994867054

멀티 카메라 시스템 내가 더블 임베딩 구성을 이용 해, 도로 이용자 이벤트(event)를 검출하기 위해서 이용되는 분할(segmentation) 성능 향상을 위한 학습 방법 및 학습 장치, 그리고 이것을 이용 했다고 실험 방법 및 실험 장치. {LEARNING METHOD AND LEARNING DEVICE FOR IMPROVING SEGMENTATION PERFORMANCE TO BE USED FOR DETECTING ROAD USER EVENTS USING DOUBLE EMBEDDING CONFIGURATION IN MULTI-CAMERA SYSTEM AND TESTING METHOD AND TESTING DEVICE USING THE SAME}LEARNING METHOD AND LEARNING DEVICE FOR IMPROVING SEGMENTATION PERFORMANCE TO BE USED FOR DETECTING ROAD USER EVENTS BY USING DOUBLE EMBEDDING CONFIGURATION IN MULTI-CAMERA SYSTEM, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method for improving segmentation performance to be used for detecting road user events in a multi-camera system. SOLUTION: A learning method includes steps of: generating similarity embedding feature by applying similarity convolution operations to a feature outputted from a neural network; generating a similarity loss by referring to similarity between two points sampled from the similarity embedding feature, and a GT label image corresponding to the similarity; generating distance embedding feature by applying distance convolution operations to t...

G06T7/10 G06T7/00

G06K9/342 G06K9/6215 G06K9/6268 G06K9/00805 G06K9/6273 G06N3/0454 G06N3/084 G06T2207/20084 G06T2207/30261 G06T7/11 G06T7/194 G05D1/0221 G05D1/0246 G05D2201/0213 G06K9/00805 G06K9/00818 G06K9/6256 G06N3/08 G06N5/046 G06T2207/30261 G06T7/194 G06T7/70 G06N20/00

G06T7/10 G06T7/00350C

5L096FA01 5L096FA32 5L096FA33 5L096FA66 5L096FA69 5L096HA11 5L096KA04 5L096KA15

31212162 2019212162

32119505 2020119505

2021.04.28

2019.11.25

2020.08.06

2021.04.12

STRADVISION INC

06865364 1994865364

에지 로스를 이용 해 보행자 이벤트(event), 자동차 이벤트(event), 폴링 이벤트(event), 포룬이벤트를 포함하는 이벤트(event)를 검출하는에 있어서 이용하시는 분할(segmentation) 성능 향상을 위한 학습 방법 및 학습 장치 및 그것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR IMPROVING SEGMENTATION PERFORMANCE TO BE USED FOR DETECTING EVENTS INCLUDING PEDESTRIAN EVENT, AUTOMOBILE EVENT, FALLING EVENT, AND FALLEN EVENT BY USING EDGE LOSS, AND TEST METHOD AND TEST DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method for improving segmentation performance to be used for detecting events such as a pedestrian event, an automobile event, a falling event, and a fallen event. SOLUTION: A learning method for improving segmentation performance includes steps of: allowing k convolutional layers to generate k encoded feature maps; allowing k-1 deconvolutional layers to sequentially generate k-1 decoded feature maps; allowing h mask layers to refer to h basic decoded feature maps outputted from h deconvolutional layers corresponding thereto and h edge feature maps ...

G06T7/00 G06T7/12

G06K9/34 G06K9/4628 G06K9/6256 G06N3/0454 G06N3/084 G06K9/4628 G06K9/6273 G06N3/0454 G06N3/084 G06T2207/20081 G06T2207/20084 G06T2207/30252 G06T7/12 G06K9/6232 G06K9/6256 G06K9/6262 G06N3/08 G06T2207/20081 G06T2207/30252 G06T7/12 G06T7/13 G06N20/00

G06T7/00350C G06T7/12

5L096EA02 5L096FA02 5L096FA06 5L096HA11 5L096KA04 5L096KA15

31202412 2019202412

32119501 2020119501

2021.04.28

2019.11.07

2020.08.06

2021.04.08

STRADVISION INC

06865366 1994865366

사람이 상태 인식을 기반으로서 신체 부위가 길이 및 얼굴 정보를 사용 해 승객이 신장 및 체중을 예측하는 방법 및 장치METHOD AND DEVICE FOR PREDICTING BODY HEIGHT AND WEIGHT OF PASSENGER USING LENGTH OF EACH BODY REGION AND FACE INFORMATION BASED ON HUMAN STATE RECOGNITION

PROBLEM TO BE SOLVED: To provide a method for detecting physical information relating to vehicle passengers based on human state recognition. SOLUTION: The method includes the steps of: when the interior image of the inside of a vehicle room is acquired, detecting the face of a passenger on the basis of the interior image by a face recognition network, outputting passenger feature information that corresponds to the detected face, detecting the body of the passenger from the interior image by a body recognition network, and outputting body region length information of the body; and searching ...

G06T7/00 G06T7/62

A61B5/1072 A61B5/1079 G01G19/44 G06K9/00255 G06K9/00268 G06K9/00288 G06K9/00369 G06K9/00838 G06K2009/00322 G06N3/0454 G06N3/084 G06K9/00268 G06K9/00288 G06K9/00362 G06K9/00832 G06K9/00838 G06K2009/00322 G06T7/60 G06T2207/20081 G06T2207/20084 G06T2207/30196 G06T2207/30201 G06T2207/30268 G06K9/00228 G06K9/00288 G06K9/00362 G06K9/00832 G06K9/6232 G06K9/6262 G06T2207/20076 G06T2207/20081 G06T2207/30201 G06T2207/30268 G06T7/62

G06T7/00660B G06T7/00350C G06T7/62

5L096BA04 5L096FA14 5L096FA34 5L096FA64 5L096HA11 5L096JA11 5L096KA04

32009532 2020009532

32123352 2020123352

2021.04.28

2020.01.23

2020.08.13

2021.04.08

STRADVISION INC

06865363 1994865363

라플라시안 피라미드 네트워크(network)를 이용 해 자율 주행(autonomous travel) 자동차 레벨 4 및 레벨 5를 만족 시키기 위해서 요구하시는 도로 장애물 검출에서 분할(segmentation) 성능 향상을 위한 학습 방법 및 학습 장치 및 이것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR IMPROVING SEGMENTATION PERFORMANCE IN ROAD OBSTACLE DETECTION REQUIRED TO SATISFY LEVEL 4 AND LEVEL 5 OF AUTONOMOUS VEHICLES BY USING LAPLACIAN PYRAMID NETWORK, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method for improving segmentation performance in detecting edges of road obstacles and traffic signs, etc. required to satisfy level 4 and level 5 of autonomous vehicles by using a learning device. SOLUTION: A learning method includes steps of: via a learning device, instructing k convolutional layers to generate k encoded feature maps, including h encoded feature maps corresponding to h mask layers; and instructing k deconvolutional layers to generate k decoded feature maps by using h bandpass feature maps and h decoded feature maps corresponding t...

G06T7/00 G06N3/08 G06N3/04

G06K9/00805 G06K9/00818 G06K9/4619 G06K9/6268 G06N3/0454 G06N3/084 G06K9/00791 G06K9/4628 G06K9/6274 G06N3/0454 G06N3/084 G06T2207/20016 G06T2207/20081 G06T2207/20084 G06T2207/30252 G06T7/11

G06T7/00350C G06N3/08 G06N3/04

5L096FA06 5L096GA10 5L096GA51 5L096HA11 5L096KA04

31202411 2019202411

32119500 2020119500

2021.04.28

2019.11.07

2020.08.06

2021.04.08

STRADVISION INC

06865365 1994865365

딥 러닝 기반 장치가 학습으로 사용하시는 가상 세계 시뮬레이터(simulator)가 물리 엔진을 캘리브래이션(calibration)하는 방법 및 장치, 그리고 그것을 위한 리얼 상태 네트워크(network)를 학습하는 방법 및 그것을 이용 했다고 학습 장치METHOD AND DEVICE FOR CALIBRATING PHYSICAL ENGINE OF VIRTUAL WORLD SIMULATOR USED FOR LEARNING OF DEEP LEARNING BASE DEVICE, METHOD FOR LEARNING REAL STATE NETWORK THEREFOR, AND LEARNING DEVICE UTILIZING THE SAME

PROBLEM TO BE SOLVED: To calibrate the physical engine of a virtual world simulator which is used in learning a deep learning base device. SOLUTION: The present invention includes the steps of: when virtual present frame information that corresponds to a virtual present state in a virtual environment is acquired, transferring the virtual present frame information to a deep learning base device and outputting virtual action information, transferring the virtual present frame information and the virtual action information to a physical engine and outputting virtual next frame information, trans...

G06T7/00 G06N3/08 G06N20/00

B60W30/08 B60W40/04 G06K9/00711 G06K9/00798 G06K9/00825 B60W60/00 G06N3/006 G06N3/0454 G06N3/088 G06K9/6256 G06K9/6262 G06N3/04 G06N3/08 G06F30/27 G06F2111/10

G06T7/00350C G06N3/08 G06N20/00

5L096BA04 5L096DA02 5L096HA02 5L096HA11 5L096KA04 5L096KA15

32007174 2020007174

32123338 2020123338

2021.04.28

2020.01.20

2020.08.13

2021.04.08

STRADVISION INC

06860943 1994860943

컨볼루션 연산의 무결성을 검증하기 위해서 테스트 패턴(test pattern)을 생성 해, 상기 테스트 패턴(pattern) 중에서 최적화 테스트 패턴(test pattern)을 선택 해, 극도가 상황으로 결함 허용 능력 및 흔들려로 강인한 성질을 향상 시키는 방법 및 장치METHOD AND DEVICE FOR GENERATING TEST PATTERNS AND SELECTING OPTIMIZED TEST PATTERNS FROM AMONG THE TEST PATTERNS IN ORDER TO VERIFY INTEGRITY OF CONVOLUTION OPERATIONS TO ENHANCE FAULT TOLERANCE AND FLUCTUATION ROBUSTNESS IN EXTREME SITUATIONS

PROBLEM TO BE SOLVED: To provide a method for generating one or more test patterns to verify integrity of one or more parameters of a convolutional neural network (CNN) by using at least one test pattern so that an autonomous vehicle can be driven safely from hacker threats. SOLUTION: A method includes steps of: generating test patterns by using a certain function so that saturation does not occur in a process to which convolution operations are applied after the test patterns are entered in at least one original CNN; generating each of evaluation scores of each of the test patterns by referr...

G06F21/51 G06F17/10 G06N3/02 G06F21/64

G06F7/588 G06F17/15 G06N3/0454 G06N3/08 G06F11/263 G06K9/00791 G06K9/6262 G06K9/627 G06F11/00 G06F11/26 G06F11/263

G06F21/51 G06F17/10A G06N3/02 G06F21/64

5B056BB26

32004560 2020004560

32119557 2020119557

2021.04.21

2020.01.15

2020.08.06

2021.03.31

STRADVISION INC

06856904 1994856904

물체가 스케일(scale)에 따라 모드 전환을 가능한 CNN 기반이 감시용 물체 검출기가 학습 방법 및 학습 장치, 그리고 이것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR OBJECT DETECTOR FOR SURVEILLANCE BASED ON CNN CAPABLE OF CONVERTING MODES ACCORDING TO SCALES OF OBJECTS, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method and a learning device capable of reducing the number of operations for allowing respective pooling layers having difference scales for each other to execute pooling operations. SOLUTION: A learning method includes steps of: allowing convolutional layers to output a feature map by applying convolution operations to an image and allowing an RPN to output ROIs in the image; allowing pooling layers to pool each of ROI areas on the feature map per each corresponding scales, allowing first FC layers to output second feature vectors, and allowing se...

G06K9/00771 G06K9/3233 G06K9/4671 G06N3/0454 G06N3/084 G06K9/3233 G06K9/6232 G06K9/6256 G06K9/6272 G06N3/0454 G06N3/084 G06K9/6231 G06K9/6232 G06K9/6256 G06K2209/23

G06T7/00350C

5L096EA39 5L096FA18 5L096FA69 5L096HA11 5L096JA11 5L096KA04

31239316 2019239316

32119520 2020119520

06856904

2021.04.14

2019.12.27

2020.08.06

2021.03.23

STRADVISION INC

06856854 1994856854

다수가 차량에 의하는 센서(sensor) 퓨전, V2X 통신 가능한 애플리케이션을 이용 했다고 협업 주행을 위해서 주위로 위치하는 카메라 각각에 대응하는 각각의 물체 검출기에 의해 검출하시는 물체 검출 정보를 통합하는 방법 및 장치METHOD AND DEVICE FOR MERGING OBJECT DETECTION INFORMATION DETECTED BY EACH OF OBJECT DETECTORS CORRESPONDING TO EACH CAMERA LOCATED IN SURROUNDINGS FOR COLLABORATIVE DRIVING USING SENSOR FUSION BY MULTIPLE VEHICLES AND V2X COMMUNICABLE APPLICATIONS

PROBLEM TO BE SOLVED: To provide an object detection method and device capable of acquiring correct object detection information with respect to surrounding environment. SOLUTION: A method for merging object detection information detected by object detectors, each of which corresponds to each of cameras located in surroundings to detect objects in each of images generated from each of the cameras by image analysis based on deep learning is provided. The method includes a step of, if first to n-th object detection information are respectively acquired from a first to an n-th object detector in...

G06K9/00771 G06K9/6202 G06K9/6268 G08G1/096725 H04N7/181 B60R1/00 B60R2300/50 G06K9/00805 G06K9/6256 G06K9/6292 H04W4/40 B60R1/00 B60R2300/303 G05D1/0246 G05D1/0276 G05D2201/0213 G06K9/00805 G06T2207/20221 G06T2207/30261 G06T7/32 H04W4/40

G06T7/00350C

5L096CA05 5L096DA01 5L096GA34 5L096HA11 5L096JA03 5L096JA16 5L096KA01 5L096KA04

32007172 2020007172

32123336 2020123336

06856854

2021.04.14

2020.01.20

2020.08.13

2021.03.23

STRADVISION INC

06857369 1994857369

CNN를 학습하는 방법 및 학습 장치, 그것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR CNN AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for learning parameters of a CNN by using a 1�K convolution operation or a K�1 convolution operation to be provided to be used for hardware optimization which satisfies a key performance index (KPI). SOLUTION: A method includes steps of: allowing a reshaping layer to two-dimensionally concatenate features in each group comprised of K channels of a training image or its processed feature map, to thereby generate a reshaped feature map; allowing a subsequent convolutional layer to apply a 1�K convolution operation or a K�1 convolution operation to the r...

G06T7/00 G06N3/04 G06N3/08

G06K9/4671 G06K9/6268 G06N3/0454 G06N3/084 G06K9/4628 G06K9/627 G06F11/3466 G06F13/378 G06F17/15 G06N3/04 G06N3/0472 G06N3/084

G06T7/00350C G06N3/04 G06N3/08 G06N3/08140

5L096FA81 5L096KA04

32001966 2020001966

32119535 2020119535

2021.04.14

2020.01.09

2020.08.06

2021.03.24

STRADVISION INC

06863619 1994863619

콘보류션뉴라르넷트워크로부터 복수가 출력이 앙상블(ensemble)을 이용 해 통합하셨다고 특징 맵을 제공하기 위한 방법 및 장치{METHOD AND DEVICE FOR PROVIDING INTEGRATED FEATURE MAP USING ENSEMBLE OF MULTIPLE OUTPUTS FROM CONVOLUTIONAL NEURAL NETWORK}METHOD AND DEVICE FOR PROVIDING INTEGRATED FEATURE MAP BY USING ENSEMBLE OF A PLURALITY OF OUTPUTS FROM CONVOLUTIONAL NEURAL NETWORK

PROBLEM TO BE SOLVED: To provide a method and a device for providing an integrated feature map by using an ensemble of a plurality of outputs from a convolutional neural network. SOLUTION: A method for acquiring an integrated feature map includes steps of: a CNN device receiving an input image and applying a plurality of deformation functions to deform the input image to generate a plurality of deformed input images; the CNN device applying convolution operation to each of the deformed input images to obtain each of deformed feature maps corresponding to each of the deformed input images; the...

G06T7/00 G06N20/20 G06N3/08 G06T1/40 G06T3/00

G06K9/00624 G06K9/6256 G06K9/6262 G06K9/6271 G06N3/0454 G06N3/08 G06N3/084 G06K9/00624 G06K9/6271 G06N3/0454 G06N3/084 G06K9/6256 G06K9/6262 G06N3/04 G06N3/08

G06T7/00350C G06N20/20 G06N3/08140 G06T1/40 G06T3/00705

5B057CD11 5B057DA17 5B057DC33 5B057DC36 5B057DC40 5L096FA25 5L096FA26 5L096FA69 5L096FA81 5L096GA34 5L096HA07 5L096HA11 5L096JA11 5L096KA15 5L096MA03

31160238 2019160238

32038664 2020038664

2021.04.21

2019.09.03

2020.03.12

2021.04.05

STRADVISION INC

06856906 1994856906

자율 주행(autonomous travel) 차량 레벨 4를 채우기 위해서 영역이 클래스에 따라 모드를 바꾸기 위해서 그리드(grid) 생성기를 이용하는 뉴럴네트워크 연산 방법 및 이것을 이용 했다고 장치NEURAL NETWORK OPERATION METHOD USING GRID GENERATOR FOR CONVERTING MODES ACCORDING TO CLASSES OF AREAS TO SATISFY LEVEL 4 OF AUTONOMOUS VEHICLES, AND DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a neural network operation method using a grid generator for converting modes according to classes of areas to satisfy level 4 of autonomous vehicles. SOLUTION: A neural network operation method includes steps of: a computing device (a) causing a pair detector to acquire information on classes and locations of pairs for testing by detecting the pairs for testing; (b) causing the grid generator to generate section information by referring to the information on locations of the pairs for testing; (c) causing a neural network to determine parameters for testing b...

G06T7/00 G06N3/08

G06K9/00791 G06K9/00805 G06K9/6256 G06K9/6261 G06K9/6262 G06K9/6274 G06K9/628 G06N3/0454 G06N3/08 G06T2207/20081 G06T2207/20084 G06T2207/30252 G06T7/70 G06K9/00805 G06K9/6274 G06N3/0454 G06N3/08 G06T2207/20081 G06T2207/20084 G06T2207/30252 G06T7/70 G06K9/00791 G06K9/6256 G06K9/6261 G06K9/6262 G06K9/628 G06N3/04 G06N3/08 G06T2207/20081 G06T2207/20084 G06T2207/30252 G06T7/70

G06T7/00350C G06N3/08

5L096AA06 5L096BA04 5L096CA04 5L096DA02 5L096FA59 5L096FA69 5L096GA19 5L096HA11 5L096JA09 5L096KA04 5L096KA15

32002012 2020002012

32119537 2020119537

06856906

2021.04.14

2020.01.09

2020.08.06

2021.03.23

STRADVISION INC

06856905 1994856905

자율 주행(autonomous travel) 차량 레벨 4를 채우기 위해서 영역이 클래스에 따라 모드를 바꾸기 위해서 그리드(grid) 생성기를 이용하는 뉴럴네트워크 연산 방법 및 이것을 이용 했다고 장치NEURAL NETWORK OPERATION METHOD USING GRID GENERATOR FOR CONVERTING MODES ACCORDING TO CLASSES OF AREAS TO SATISFY LEVEL 4 OF AUTONOMOUS VEHICLES, AND DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a neural network operation method using a grid generator for converting modes according to classes of areas to satisfy level 4 of autonomous vehicles. SOLUTION: A neural network operation method includes steps of: a computing device (a) causing a detector to acquire object location information for testing and class information; (b) causing the grid generator to generate section information by referring to the object location information for testing; (c) causing a neural network to determine parameters for testing, to be used for applying the neural network ope...

G06T7/00 G06N3/08

G06K9/00791 G06K9/342 G06K9/6256 G06N3/02 G06K9/00791 G06K9/6267 G06N3/0454 G06N3/08 G06K9/00791 G06K9/6257 G06K9/6261 G06K9/6262 G06K9/628 G06N3/04 G06N3/08 G06N7/005 G06T2207/30252 G06T7/70

G06T7/00350C G06N3/08

5L096AA06 5L096BA04 5L096CA04 5L096DA02 5L096FA69 5L096GA19 5L096HA11 5L096KA04 5L096KA15

32001999 2020001999

32119536 2020119536

06856905

2021.04.14

2020.01.09

2020.08.06

2021.03.23

STRADVISION INC

06856957 1994856957

사진 스타일(style) 변환을 사용 해 런타임(run time) 입력이 변환에 따라 리얼 세계로 사용하시는 가상 세계로 학습하셨다고 CNN를 허용하기 위한 학습 방법 및 학습 장치, 그리고 이것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR ALLOWING CNN HAVING TRAINED IN VIRTUAL WORLD TO BE USED IN REAL WORLD BY RUNTIME INPUT TRANSFORMATION USING PHOTO STYLE TRANSFORMATION, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To allow a CNN having trained in a virtual world to be used in a real world. SOLUTION: A leaning method includes steps of: a learning device acquiring first learning images of virtual driving of a virtual vehicle; and the learning device performing a first learning process of instructing a main CNN to generate first estimated autonomous driving source information by referencing the first learning images, instructing the main CNN to generate first main losses by referencing first ground-truth autonomous driving source information corresponding to a first estimation and pe...

G06N3/08 G06T7/00

G06K9/00791 G06K9/6256 G06N3/0454 G06N3/084 G06T3/0012 G06K9/00791 G06K9/4628 G06K9/6255 G06K9/6256 G06K9/627 G06N3/0418 G06N3/0454 G06N3/084 G05D1/0088 G05D2201/0213 G06K9/00791 G06K9/6256 G06N3/0418 G06N3/0454 G06N3/084

G06N3/08 G06T7/00350C G06T7/00650Z

5L096BA04 5L096CA18 5L096HA11 5L096KA04

32008618 2020008618

32123344 2020123344

2021.04.14

2020.01.22

2020.08.13

2021.03.23

STRADVISION INC

06856853 1994856853

극한 상황에 있어서 폴트 톨러런스 및 프랙션 츄 A 숀 로버스트 성을 향상 시키기 위해서, 짓타링을 일어났다고 이미지를 안정화 시키는 프로세스(process)로 생성하셨다고 warping하셨다고 이미지로 발생 했다고 왜곡을, GAN를 이용 해 줄이기 위한 학습 방법 및 학습 장치, 그리고 그것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR REDUCING DISTORTION OCCURRED IN WARPED IMAGE GENERATED IN PROCESS OF STABILIZING JITTERED IMAGE BY USING GAN TO ENHANCE FAULT TOLERANCE AND FLUCTUATION ROBUSTNESS IN EXTREME SITUATIONS, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method for reducing distortion occurred in a warped image by using a GAN for enhancing fault tolerance and fluctuation robustness in extreme situations. SOLUTION: A learning method includes steps of: if an initial image is acquired, allowing an adjusting layer included in a generating network to adjust at least part of initial feature values, to thereby transform the initial image into an adjusted image; and if at least part of a naturality score, a property retention score, and a similarity score are acquired, allowing a loss layer included in the ...

G06T7/00 G06N3/04 G06N3/08

G06T5/006 G06T2207/10004 G06T2207/20081 G06T2207/20084 G06T2207/30248 G06K9/6261 G06N3/0454 G06N3/0472 G06N3/084 G06N3/088 G06T5/003 G06T5/006 G06T2207/20081 G06T2207/20084 G06T2207/20201 G06T2207/30252 H04N5/23267 G06K9/6215 G06K9/6256 G06K9/6262 G06K9/628 G06N3/04 G06N3/084 G06N3/088 G06T5/006 G06T7/0002 G06T2207/20076 G06T2207/20081 G06T2207/20084 G06T2207/30252 H04N5/23264

G06T7/00350C G06N3/04154 G06N3/08180

5L096HA11 5L096JA03 5L096JA11 5L096KA04 5L096MA03

32004558 2020004558

32119555 2020119555

2021.04.14

2020.01.15

2020.08.06

2021.03.23

STRADVISION INC

06856852 1994856852

가로 필터(filter) 마스크를 이용 해 자율 주행(autonomous travel) 차의 주행 경로를 계획하기 위해서 차선 요소를 검출하는 방법 및 장치METHOD AND DEVICE FOR DETECTING LANE ELEMENT IN ORDER TO PLAN TRAVEL ROUTE OF AUTONOMOUS VEHICLE UTILIZING HORIZONTAL FILTER MASK

PROBLEM TO BE SOLVED: To provide a method for detecting a lane element in order to plan the travel route of an autonomous vehicle. SOLUTION: The method for detecting a lane element includes: a step S01 for a CNN to generate a segmentation score map utilizing an input image; a step S02 for causing a post-processing module that carries out data processing at the output end of the CNN to generate a magnitude map utilizing the segmentation score map and a horizontal filter mask; a step S03 for causing each of lane element candidate groups per row of the segmentation score map to be determined ref...

G06T7/60 G06T7/00 G08G1/16

G06K9/00798 G06N3/0454 G06T5/002 G06T7/11 G06T7/136 G06K9/00798 G06K9/34 G06K9/4609 G06K9/4628 G06T2207/20084 G06T2207/30256 G08G1/167 G06T7/11 G06T7/73 G06K9/00798 G06T2207/10016 G06T2207/20084 G06T2207/30256 G08G1/04 G08G1/167 G06T7/73

G06T7/60200J G06T7/00650A G06T7/00350C G08G1/16C

5H181AA01 5H181AA26 5H181AA27 5H181BB15 5H181BB20 5H181CC24 5H181FF27 5H181LL09 5L096BA04 5L096CA02 5L096FA14 5L096GA55 5L096HA11

32004555 2020004555

32126609 2020126609

2021.04.14

2020.01.15

2020.08.20

2021.03.23

STRADVISION INC

06856952 1994856952

복수가 비디오 프레임(frame)을 이용 해 CNN가 파라미터(parameter)를 최적화하기 위한 학습 방법 및 학습 장치, 그리고 이것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR OPTIMIZING PARAMETER OF CNN BY USING A PLURALITY OF VIDEO FRAMES, AND TEST METHOD AND TEST DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method and a learning device for optimizing a parameter of a CNN by using a plurality of video frames, and a test method and a test device using the same. SOLUTION: A learning method includes the steps of: a CNN learning device applying at least one convolutional operation respectively to a t-k-th input image corresponding to a t-k-th frame and a t-th input image corresponding to a t-th frame being a frame following the t-k-th frame as training images, and thereby obtaining a t-k-th feature map corresponding to the t-k-th frame and a t-th feature ma...

G06K9/00362 G06K9/00744 G06K9/00758 G06N3/0454 G06N3/084 G06K9/00744 G06K9/3241 G06K9/34 G06K9/4628 G06N3/0454 G06N3/084 G06F17/16 G06K9/6202 G06K9/6262 G06N3/084

G06T7/00350C

5L096EA39 5L096FA66 5L096HA11 5L096KA04

31160651 2019160651

32038669 2020038669

2021.04.14

2019.09.03

2020.03.12

2021.03.23

STRADVISION INC

06856851 1994856851

이미지 인코딩(encoding)을 위한 방법 및 장치 그리고 이것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)METHOD AND DEVICE FOR IMAGE ENCODING, AND TEST METHOD AND TEST DEVICE UTILIZING THE SAME

PROBLEM TO BE SOLVED: To provide an encoding method which allows features of more diverse properties to be extracted by taking into consideration various kernel sizes, in a CNN. SOLUTION: An image encoding method includes the steps, that are performed by a learning device comprising first to n-th convolution layers, of: obtaining an input image; applying transposed convolution calculation with respect to the input image or an input feature map obtained previously from the convolution layers, by using a part of the convolution layers, and creating transposed feature maps having different sizes...

G06N3/084 G06T2207/20081 G06T2207/20084 G06T7/10 G06T7/11 G06N3/0454 G06N3/0481 G06N3/084 G06T9/002 G06T2207/20081 G06T2207/20084 G06T7/11 H04N19/90 G06N3/084 G06T9/002 G06T2207/20081 G06T2207/20084 G06T7/10

31163898 2019163898

32047271 2020047271

06856851

2021.04.14

2019.09.09

2020.03.26

2021.03.23

STRADVISION INC

06856855 1994856855

그 자체에 의해 생성하셨다고 정보 및 다른 개체에 의해 생성하셨다고 정보를 선택적으로 이용 해 카메라가 오정렬을 보정하는 방법 및 이것을 이용 했다고 장치METHOD FOR CORRECTING WRONG ARRANGEMENT OF CAMERAS BY SELECTIVELY USING INFORMATION GENERATED BY ITSELF AND INFORMATION GENERATED BY OTHER OBJECTS AND DEVICE FOR USING THE SAME

PROBLEM TO BE SOLVED: To provide a method and a computing device for correcting the angle of dislocation of a camera. SOLUTION: The method includes the steps in which: a computing device generates first reference data and second reference data by circumstantial information with reference to reference images; the computing device acquires a first angle error or a second angle error with reference to one of the first reference data and the second reference data and with reference to vehicle coordinate data; and the computing device adjusts the angel of dislocation by a physical rotation module ...

G06T7/80 B60W40/02 G06T7/00 G08G1/09

G06K9/00798 G06K9/00825 G06N3/0454 H04N17/002 G06K9/00798 G06K9/00832 G06K9/03 G06K9/3216 G06K9/4628 G06K2209/23 G06N3/0454 G06T2207/20084 G06T2207/30242 G06T2207/30256 G06T2207/30261 G06T7/80 H04N17/002 H04N5/23299 G06N3/0454 G06T2207/20084 G06T2207/30256 G06T7/74 H04N5/23299

G06T7/80 B60W40/02 G06T7/00350C G06T7/00650Z G08G1/09H

3D241BA63 3D241CE03 3D241CE06 3D241DC01Z 3D241DC02B 3D241DC02Z 3D241DC21Z 3D241DC22B 3D241DC22Z 3D241DC25Z 3D241DC35Z 3D241DC37Z 5H181AA01 5H181BB04 5H181BB17 5H181BB18 5H181CC04 5H181FF13 5H181FF27 5L096AA06 5L096BA04 5L096BA18 5L096CA02 5L096DA02 5L096EA39 5L096FA32 5L096FA66 5L096FA67 5L096FA69 5L096FA76 5L096GA51 5L096HA11 5L096JA11

32011164 2020011164

32126647 2020126647

06856855

2021.04.14

2020.01.27

2020.08.20

2021.03.23

STRADVISION INC

06857370 1994857370

타겟(destination) 물체 예측 네트워크(network) 및 타겟(destination) 물체 통합 네트워크(network)를 이용 해, 중요 실적평가 지표와 같은 사용자(user) 요구 사항과 관련된 최적화를 위한 재구성 가능한 네트워크(network) 기반이 물체 검출기를 학습하는 방법 및 학습 장치 및 이것을 이용 했다고 실험 방법 및 실험 장치METHOD AND DEVICE FOR LEARNING NETWORK-BASED OBJECT DETECTOR WHICH CAN BE RECONFIGURED FOR OPTIMIZATION RELATING TO USER`S REQUIREMENT SUCH AS IMPORTANT PERFORMANCE EVALUATION INDEX USING TARGET OBJECT PREDICTION NETWORK AND TARGET OBJECT INTEGRATED NETWORK, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for learning parameters of a CNN-based object detector optimal to user's requirements such as an important performance evaluation index using a target object prediction network and an object integrated network, which improves accuracy of a 2D bounding box and is usefully carried out for a multiple-camera, a surround view monitoring, etc. SOLUTION: A CNN is redesigned according to change in scale of an object due to change of a resolution or focal distance relating to an important performance evaluation index. The method includes the steps in which: a ...

G06T7/00 G06N3/04

G06K9/3233 G06K2209/21 G06N3/0454 G06N3/084 G06T2207/10004 G06T7/62 G06K9/00805 G06K9/3233 G06K9/4628 G06K9/627 G06N3/0454 G06N3/084 G06K9/6228 G06K9/6232 G06K9/6256 G06K9/6262 G06K9/6288 G06N3/08

G06T7/00350C G06N3/04154

5L096BA04 5L096FA19 5L096HA03 5L096KA04

32002986 2020002986

32119545 2020119545

2021.04.14

2020.01.10

2020.08.06

2021.03.24

STRADVISION INC

06858382 1994858382

하드웨어(hardware) 최적화로 사용하시는 1 xH콘보류션을 이용 했다고 CNN 기반이 물체 검출기를 학습하는 방법 및 학습 장치, 그것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)METHOD AND LEARNING DEVICE FOR LEARNING OBJECT DETECTOR OF CNN FOUNDATION USING 1��H CONVOLUTION USED FOR HARDWARE OPTIMIZATION, TEST METHOD AND TEST DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for learning a parameter of an object detector of a CNN foundation. SOLUTION: A method of the invention comprises the steps of: using a first transpose layer or a pooling layer, for concatenating pixels on a pooled map for every proposal for every proposal; using 1�H1 and 1�H2 convolution layers, for applying 1�H1 and 1�H2 convolution calculations with respect to a reshaped map which is generated by concatenating features in corresponding channels out of all channels of a concatenated pooled map, for generating an adjustment map; and using a second tr...

G06T7/00 G06N3/04

G06K9/6256 G06K9/6262 G06N3/0454 G06K9/00624 G06K9/4628 G06K9/6274 G06N3/0454 G06N3/084 G06K9/42 G06K9/46 G06K9/6232 G06K9/6256 G06K9/6261 G06K9/6262 G06K9/6267 G06N3/08

G06T7/00350C G06N3/04

5L096CA01 5L096DA02 5L096FA69 5L096GA55 5L096HA11 5L096JA11 5L096KA04 5L096KA15

32002054 2020002054

32119539 2020119539

2021.04.14

2020.01.09

2020.08.06

2021.03.26

STRADVISION INC

06857368 1994857368

CNN 학습 방법 및 학습 장치, 이것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR CNN, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method of parameters of a CNN for image recognition to be used for hardware optimization which satisfies a key performance index (KPI). SOLUTION: A method includes steps of: allowing a first transposing layer or a pooling layer to generate an integrated feature map by concatenating pixels, per each ROI, on pooled ROI feature maps; allowing an 1xH1 convolutional layer to generate a first adjusted feature map by using a first reshaped feature map, generated by concatenating features in H1 channels of the integrated feature map; allowing an 1xH2 convol...

G06T7/00 G06T7/11 G06N3/04

G06K9/3233 G06K9/6267 G06N3/0454 G06N3/084 G06K9/3241 G06K9/4628 G06K9/6232 G06K9/6268 G06T2207/20081 G06T2207/20084 G06T7/11

G06T7/00350C G06T7/11 G06N3/04

5L096AA06 5L096EA03 5L096EA39 5L096FA52 5L096FA64 5L096GA55 5L096HA11 5L096JA11 5L096KA04 5L096MA07

32001957 2020001957

32119534 2020119534

2021.04.14

2020.01.09

2020.08.06

2021.03.24

STRADVISION INC

06857371 1994857371

안전한 자율 주행(autonomous travel)을 지원하는 학습 방법, 그리고 이것을 이용 했다고 학습 장치, 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD FOR SUPPORTING SAFE AUTONOMOUS DRIVING, LEARNING DEVICE, TESTING METHOD, AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method for supporting execution of safe autonomous driving through a fusion of information acquired from images and via communication. SOLUTION: In a learning method, a learning device is configured to: allow a first neural network and a second neural network to generate an image-based feature map and a communication-based feature map by using a circumstance image and circumstance communication information; allow a third neural network to apply a third neural network operation to the image-based feature map and the communication-based feature map to...

G06T1/00 G08G1/16 G06N3/04 G06N3/08

G06K9/00791 G06N3/0454 G06N3/084 G06K9/00791 G06K9/00805 G06K9/629 B60W50/00 B60W2050/0004 B60W2420/42 B60W60/00 G05D1/0088 G06K9/6262 G06K9/6289 G06N3/0454 G06N3/084 G06N20/10 G06N20/20

G06T1/00330 G08G1/16A G06N3/04154 G06N3/08140 G06N3/04145

5B057BA01 5B057CA08 5B057CA12 5B057DB02 5B057DB09 5B057DC01 5H181AA01 5H181BB04 5H181CC04 5H181LL01 5H181LL02 5H181LL04 5H181LL09

32010125 2020010125

32126639 2020126639

2021.04.14

2020.01.24

2020.08.20

2021.03.24

STRADVISION INC

06855082 1994855082

자율 주행(autonomous travel) 상황에서의 장애물 검출을 위한 CNN 학습용무 이미지 데이터(image data) 세트가 생성 방법 및 장치, 그리고 이것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)METHOD AND DEVICE FOR GENERATING IMAGE DATA SET FOR CNN LEARNING FOR DETECTION OF OBSTACLE IN AUTONOMOUS DRIVING CIRCUMSTANCES AND TEST METHOD AND TEST DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method and a device for generating an image data set for CNN (convolution neural network) learning for ensuring an image set for learning used for improving performance to detect an object not to be frequently in contact in actual road driving circumstances in autonomous driving, and a test method and a test device using the same. SOLUTION: A method for generating an image data set for CNN learning for detecting an obstacle in autonomous driving circumstances includes the steps in which a learning device: acquires an original image indicating road driving ci...

G06K9/00805 G06K9/6268 G06N3/0454 G06N3/084 G06K9/00664 G06K9/00805 G06K9/4628 G06K9/6256 G06K9/6274 G06N3/0454 G06N3/084 G06N7/005 G05D1/0088 G05D2201/0213 G06K9/00805 G06K9/6257 G06N3/0454 G06N3/084

G06T7/00350C G06T7/00650A

5L096AA02 5L096BA04 5L096DA01 5L096DA02 5L096FA18 5L096FA69 5L096GA10 5L096HA11 5L096JA11 5L096KA04 5L096KA15

31160649 2019160649

32038667 2020038667

2021.04.07

2019.09.03

2020.03.12

2021.03.19

STRADVISION INC

06849888 1994849888

주변 차량이 이미지를 이용 해 베이스 차량이 이미지를 오토 레이블링(labelling)하는 오토 레이블링(labelling) 장치를 학습 시키는 학습 방법 및 학습 장치, 그리고, 이것을 이용 했다고 실험 방법 및 실험 장치LEARNING METHOD AND LEARNING DEVICE FOR LEARNING AUTOMATIC LABELING DEVICE CAPABLE OF AUTO-LABELING IMAGE OF BASE VEHICLE USING IMAGES OF NEARBY VEHICLES, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for learning a device capable of auto-labeling a base image of a base vehicle using a sub-image of a nearby vehicle. SOLUTION: A method for learning an automatic labeling device for auto-labeling an image of a base vehicle using sub-images of nearby vehicles includes the steps of: generating a dense correspondence from a base image and sub-images; inputting the base image and the sub-images, outputting a convolution feature map, and outputting a deconvolution feature map from the convolution feature map; moving a coordinate of a (k+1)-th deconvolution...

G06T7/00 G06N20/00

G06N3/0454 G06N3/084 G06T2207/20081 G06T7/10 G06N20/00 G06K9/00791 G06K9/4628 G06K9/627 G06T2207/20081 G06T2207/20084 G06T2207/30252 G06T7/11 G06T7/174 G06K9/6259 G06K9/6264 G06N3/04 G06N3/084

G06T7/00650B G06T7/00350C G06N20/00130

5L096BA04 5L096EA39 5L096FA69 5L096GA10 5L096GA34 5L096HA11 5L096KA04

32009847 2020009847

32126635 2020126635

06849888

2021.03.31

2020.01.24

2020.08.20

2021.03.09

STRADVISION INC

06855091 1994855091

뉴럴네트워크학습으로 이용하시는 오토 레이블링(labelling)하셨다고 이미지중에서 라벨 검수를 위한 샘플(sample) 이미지를 취득하는 방법 및 그것을 이용 했다고 샘플(sample) 이미지 획득 장치METHOD FOR ACQUIRING SAMPLE IMAGE FOR LABEL ACCEPTANCE INSPECTION FROM AMONG AUTO-LABELED IMAGES UTILIZED FOR NEURAL NETWORK LEARNING, AND SAMPLE IMAGE ACQUISITION DEVICE UTILIZING THE SAME

PROBLEM TO BE SOLVED: To select an image for acceptance inspection, from among images auto-labeled by an auto-labeling device, which is difficult to label and reduce the time needed for acceptance inspection. SOLUTION: Provided is a method for acquiring a sample image for label acceptance inspection from auto-labeled images used for learning a network through deep learning, the method comprising: a step of generating a concatenated feature map by a sample image acquisition device 100 generating a first image and a second image, a convolution layer generating a first feature map and a second f...

G06T7/00 G06N3/08

G06K9/6256 G06N3/0454 G06N3/084 G06K9/00664 G06K9/6262 G06K9/6271 G06N3/0454 G06N3/084 G06K9/6232 G06K9/6257 G06K9/6262 G06K9/6277 G06K9/628 G06N3/04 G06N3/084

G06T7/00350C G06N3/08

5L096GA30 5L096GA34 5L096HA11 5L096KA04

32004613 2020004613

32123330 2020123330

06855091

2021.04.07

2020.01.15

2020.08.13

2021.03.19

STRADVISION INC

06855090 1994855090

카메라로부터 취득하셨다고 이미지와, 그에 대응하는 레이더 또는 라이다를 통하고 취득하셨다고 포인트 클라우드 맵을 뉴럴네트워크가 각각의 콘보류션스테이지마다 통합하는 학습 방법 및 학습 장치, 그리고 그것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE FOR INTEGRATING IMAGE ACQUIRED FROM CAMERA AND POINT CLOUD MAP ACQUIRED THROUGH CORRESPONDING RADAR OR LIDAR FOR EACH CONVOLUTION STAGE OF NEURAL NETWORK, AND TEST METHOD AND TEST DEVICE UTILIZING THE SAME

PROBLEM TO BE SOLVED: To provide an efficient method for integrating an image by a camera and a point cloud map by the corresponding radar or lidar for each convolution stage of a neural network. SOLUTION: The method includes: an initial computation layer adding depth information included in an original point cloud map to an original image and generating a first fused feature map 201_1 and adding hue information included in the image to the map and generating a first fused point cloud map 202_1; a conversion layer computing first conversion of the first fused feature map and generating a 1_1t...

G06N3/04 G01S13/86 G01S13/931 G01S17/87 G01S17/93 G06T7/00 G05D1/02

G06N3/0454 G06T5/50 G06T2207/10028 G06T2207/10044 G06T2207/20084 G06T2207/20221 G01S13/867 G01S13/89 G01S13/931 G01S17/89 G06K9/00791 G06K9/4628 G06K9/6271 G06K9/6288 G06N3/0454 G06N3/08 G01S17/86 G01S17/931 G01S17/89 G06K9/00201 G06K9/6256 G06K9/66 G06N3/08 G06T2207/10028 G06T7/521

G06N3/04 G01S13/86 G01S13/931 G01S17/87 G01S17/93 G06T7/00350C G05D1/02K

5H301AA01 5H301AA10 5H301CC03 5H301CC06 5H301CC10 5H301GG08 5H301GG09 5J070AC02 5J070AC11 5J070AF03 5J070BD08 5J084AA05 5J084AA10 5J084AC02 5L096BA04 5L096CA02 5L096FA66 5L096HA11 5L096JA11 5L096KA04 5L096KA15

32004226 2020004226

32126607 2020126607

06855090

2021.04.07

2020.01.15

2020.08.20

2021.03.19

STRADVISION INC

06846069 1994846069

원거리 검출 또는 군사 목적을 위해서, 이메이지콘카치네이션을 이용 했다고 CNN 기반이 하드웨어(hardware) 최적화를 가능한 물체 검출기를 학습하는 방법 및 학습 장치, 그것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment){LEARNING METHOD AND LEARNING DEVICE FOR OBJECT DETECTOR WITH HARDWARE OPTIMIZATION BASED ON CNN FOR DETECTION AT DISTANCE OR MILITARY PURPOSE USING IMAGE CONCATENATION, AND TESTING METHOD AND TESTING DEVICE USING THE SAME}LEARNING METHOD AND LEARNING DEVICE FOR OBJECT DETECTOR CAPABLE OF HARDWARE OPTIMIZATION BASED ON CNN FOR DETECTION AT LONG DISTANCE OR MILITARY PURPOSE USING IMAGE CONCATENATION, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a method for learning an object detector capable of hardware optimization based on a CNN for detection at long distance or a military purpose by using image concatenation. SOLUTION: A CNN can be redesigned when scales of objects change as focal length or a resolution changes depending on a KPI. A method includes steps of: (a) concatenating an n-th manipulated image which corresponds to an n-th target region; (b) allowing an RPN to generate first to n-th object proposals in the n-th manipulated image by using an integrated feature map, and allowing a pooling la...

G06T7/00 G06N3/08

G06K9/00664 G06K2209/21 G06N3/0454 G06N3/084 G06K9/00805 G06K9/3233 G06K9/4628 G06K9/6256 G06K9/627 G06K9/6273 G06N3/0454 G06N3/084 G06K9/4642 G06K9/6257 G06K9/6262 G06N3/04 G06N7/005

G06T7/00350C G06N3/08140

5L096EA35 5L096HA11 5L096JA11 5L096KA04

32002303 2020002303

32119540 2020119540

06846069

2021.03.24

2020.01.09

2020.08.06

2021.03.03

STRADVISION INC

06850047 1994850047

자율주행모드(autonomous travel)와 매뉴얼(documentation) 주행 모드와의 사이의 주행 모드를 변경하기 위해서 자율 주행(autonomous travel)이 안전을 확인하기 위한 RNN가 학습 방법 및 학습 장치, 그리고 테스트 방법 및 테스트 장치(test equipment)LEARNING METHOD AND LEARNING DEVICE OF RECURRENT NEURAL NETWORK (RNN) FOR CHECKING AUTONOMOUS TRAVELING SAFETY FOR CHANGING TRAVELING MODE BETWEEN AUTONOMOUS TRAVELING MODE AND MANUAL TRAVELING MODE, AND TESTING METHOD AND TESTING DEVICE

PROBLEM TO BE SOLVED: To provide a method and a device for learning a recurrent neural network (RNN) for changing a mode between an autonomous traveling mode and a manual traveling mode. SOLUTION: By a learning device, a learning method includes steps of: if training images corresponding to front and rear cameras of an autonomous vehicle are acquired, inputting each pair of the training images into corresponding CNNs, to concatenate the training images and generate feature maps for learning; inputting the feature maps for learning into long short-term memory (LSTM) models corresponding to seq...

G06T7/00 G06N3/04 G08G1/00 B60W40/02

G06K9/00791 G06K9/6256 G06K9/6262 G06N3/0445 G06N3/0454 G06K9/00791 G06K9/6271 G05D1/0088 G05D1/0214 G05D1/0231 G05D2201/0213 G06K9/00791 G06K9/6231 G06K9/6257 G06K9/6262 G06N3/0445 G06N3/0454 G06N3/084

G06T7/00350C G06N3/04154 G08G1/00D G08G1/00X B60W40/02 G06T7/00650Z

3D241BA29 3D241BA31 3D241BB40 3D241BB52 3D241BB54 3D241BB59 3D241BC01 3D241BC02 3D241BC04 3D241CC01 3D241CC08 3D241CC17 3D241CD07 3D241CD09 3D241CD10 3D241CD12 3D241CE08 3D241DC01Z 3D241DC23Z 3D241DC24Z 3D241DC25Z 3D241DC31Z 3D241DC33Z 5H181AA03 5H181BB13 5H181BB20 5H181CC04 5H181FF10 5H181LL09 5L096BA04 5L096CA05 5L096HA07 5L096HA11 5L096JA13 5L096JA16 5L096KA04

32007151 2020007151

32123334 2020123334

2021.03.31

2020.01.20

2020.08.13

2021.03.09

STRADVISION INC

06846062 1994846062

근접 장애물이 하단 라인 및 상단 라인을 검출 해, 물체 존재성을 검출하는 학습 방법 및 학습 장치, 그리고 이것을 이용 했다고 테스트 방법 및 테스트 장치(test equipment){LEARNING METHOD, LEARNING DEVICE FOR DETECTING OBJECTNESS BY DETECTING BOTTOM LINE AND TOP LINE OF NEAREST OBSTACLES AND TESTING METHOD, TESTING DEVICE USING THE SAME}LEARNING METHOD AND LEARNING DEVICE FOR DETECTING OBJECT EXISTENCE BY DETECTING BOTTOM LINES AND TOP LINES OF NEAREST OBSTACLES, AND TESTING METHOD AND TESTING DEVICE USING THE SAME

PROBLEM TO BE SOLVED: To provide a learning method and a learning device for detecting object existence by detecting bottom lines and top lines of nearest obstacles. SOLUTION: A method is characterized in that a learning device includes: (a) a step of causing a first CNN to generate first encoded feature maps and first decoded feature maps, and causing a second CNN to generate second encoded feature maps and second decoded feature maps; (b) a step of generating, respectively, first and second obstacle segmentation results in which where bottom lines and top lines line are located for each col...

G06K9/00805 G06K9/6262 G06K9/6271 G06K9/00805 G06K9/6262 G06K9/6271 G06N3/0454 G06N3/084 G06K9/00805 G06K9/4604 G06K9/66 G06N3/084 G06N5/046

G06T7/00350C

5L096AA06 5L096BA04 5L096CA04 5L096DA02 5L096FA19 5L096HA09 5L096HA11 5L096JA11 5L096JA18 5L096KA04

31161677 2019161677

32068014 2020068014

2021.03.24

2019.09.05

2020.04.30

2021.03.03

STRADVISION INC