1 of 66

CRICKET SKILLS DEVELOPMENT

2021-172

2 of 66

Team Members

Name

IT Number

R.Vithurson

IT18142556

A.M.Karthik

IT18190762

D.Mithelan

IT18125344

A.Narthanan

IT18188196

3 of 66

Introduction

  • The proposed system will help the players to improve their cricketing skills in batting, bowling and fielding.
  • Classify various types of batting shots, accuracy of a shot, comparison between two players.
  • Identifying bowls length, bowls' speed and wicket taking ability.
  • Users will be able to find statistics based on their performance.

​

​

4 of 66

Proposed system “Cricket4U”

5 of 66

Research Question

  • Is technology Involved in domestic cricket ?
  • Do we record players performance ?

6 of 66

Objectives

Main Objective

  • To develop an application to improve the cricket skills of emerging cricketers.

​

Sub Objective

  • To identify a human body movement action with related to cricketing shots.
  • To identify a bowler’s length and the balls pitching position.
  • Comparing the captured body movements with professional cricket shots.
  • Calculating the balls wicket taking percentage according to the balls pitching length.

​

7 of 66

System Overview

8 of 66

  

R.VITHURSON  | IT18142556

​

BSc Hons information Technology specialization in Software Engineering

R.Vithurson | IT18142556 | Project id- 2021-172

9 of 66

Introduction

  • Taking Video as a user input and extracting the human pose from that Video.

​

​

​

R.Vithurson | IT18142556 | Project id- 2021-172

10 of 66

Research Questions

  • How to split the video into frames to capture the body movement?
  • What are the ways of capturing user's body movement ?
  • How to extract the coordinates from the video?

R.Vithurson | IT18142556 | Project id- 2021-172

11 of 66

Objectives

Main Objectives

  • Capturing User's Body Movement.

​

Sub-Objectives

  • Getting input video from User through Mobile Application.
  • Capturing User's body movement.
  • Extracting coordinates from captured body Movement .

​

​

​

​

R.Vithurson | IT18142556 | Project id- 2021-172

12 of 66

Methodology

  • Taking User's video as input and Converting video as frames .
  • By using Human Pose Estimation in OpenCV and Mediapipe, we can identify body movement .
  • From  captured Human body movement we can extract coordinates.
  • Coordinates will be sent to model to comparison.

​

​

R.Vithurson | IT18142556 | Project id- 2021-172

13 of 66

Technologies used

Technologies

  • OpenCV and MediaPipe
  • Mobile Application- React Native.

​

Tools

  • Version Control System – Gitlab
  • Project Management- Azure Boards

​

​

​

​

R.Vithurson | IT18142556 | Project id- 2021-172

14 of 66

System Diagram

R.Vithurson | IT18142556 | Project id- 2021-172

15 of 66

Completion of the project

R.Vithurson | IT18142556 | Project id- 2021-172

16 of 66

Completion of the project

R.Vithurson | IT18142556 | Project id- 2021-172

17 of 66

Currently Working on

R.Vithurson | IT18142556 | Project id- 2021-172

18 of 66

Gantt Chart

R.Vithurson | IT18142556 | Project id- 2021-172

19 of 66

References

[1] A. Jalal, A. Nadeem and S. Bobasu, "Human Body Parts Estimation and Detection for Physical Sports Movements," 2019 2nd International Conference on Communication, Computing and Digital systems (C-CODE), Islamabad, Pakistan, 2019, pp. 104-109, doi: 10.1109/C-CODE.2019.8680993.

[2] R. R. Bharath and G. Dhivya, "Moving object detection, classification and its parametric evaluation," International Conference on Information Communication and Embedded Systems (ICICES2014), Chennai, India, 2014, pp. 1-6, doi: 10.1109/ICICES.2014.7033891.

R.Vithurson | IT18142556 | Project id- 2021-172

20 of 66

IT18125344 | Mithelan Devanandan

BSc Hons in Information Technology Specializing in Software Engineering

Mithelan.D | IT18125344 | Project Id- 2021-172

21 of 66

Introduction 

  • Identifying the cricket shot performed by the user.
  • User is to compare their cricket shot with an international player.
  • Mobile Application for all the people who have an interest on cricket.

​

​

Mithelan.D | IT18125344 | Project Id- 2021-172

21

22 of 66

Research  Questions 

​

  • How to capture the body coordinates?
  • How to detect the Cricket shot?
  • How to compare the cricket Shots?

Mithelan.D | IT18125344 | Project Id- 2021-172

23 of 66

Objectives

Main Objectives

  • Identifying the type of cricket shot is played.
  • Comparing the cricket shot with an international player.

​

Sub-Objectives

  • Comparing the co-ordinates of the cricket shot using Random forest algorithm.
  • Obtaining the body co-ordinates from the user.
  • Assist the user

​

​

​

Mithelan.D | IT18125344 | Project Id- 2021-172

24 of 66

Methodology

  • Obtaining the images/videos of cricket shots[2].
  • Initially tried with TF-Pose estimation.
  • By using Mediapipe Pose estimation, we can get the co-ordinates.
  • Random Forest Algorithm to classify the cricket shots.

Mithelan.D | IT18125344 | Project Id- 2021-172

Trained dataset with mediapipe

25 of 66

Methodology

  • Collected the images/videos of particular international cricket players.
  • Trained 2900+ images.

Mithelan.D | IT18125344 | Project Id- 2021-172

Coverdrive dataset with 30 co-ordinates

Accuracy of random forest algorithm

26 of 66

System Diagram

Mithelan.D | IT18125344 | Project Id- 2021-172

Mithelan.D | IT18125344 | Project Id- 2021-172

27 of 66

Completion of the project

Mithelan.D | IT18125344 | Project Id- 2021-172

28 of 66

Dataset- Cut shot & Cover Drive

Mithelan.D | IT18125344 | Project Id- 2021-172

29 of 66

Dataset- Pull shot & Straight Shot

Mithelan.D | IT18125344 | Project Id- 2021-172

30 of 66

Mithelan.D | IT18125344 | Project Id- 2021-172

API- fastAPI

31 of 66

Prototypes

Mithelan.D | IT18125344 | Project Id- 2021-172

32 of 66

Integrated System

Mithelan.D | IT18125344 | Project Id- 2021-172

33 of 66

Technologies used

Technologies

  • Capturing key points of body- MediaPipe
  • For classifying the shots – Random forest Algorithm
  • Version Control System – GIT
  • Project Management- Microsoft Teams

​

Mobile Application

  • Frontend- React-native
  • Backend – Python and FastAPI

​

​

​

​

​

Mithelan.D | IT18125344 | Project Id- 2021-172

34 of 66

Future Work

  • There will be two features for the user to classify their shots.
  • Classfiying the cricket shot using LSTM Algorithm.

​

​

Mithelan.D | IT18125344 | Project Id- 2021-172

35 of 66

Gantt Chart

Mithelan.D | IT18125344 | Project Id- 2021-172

Future work

36 of 66

References

[1] M. Zeeshan Khan, M. A. Hassan, A. Farooq and M. U. Ghanni Khan, "Deep CNN Based Data-Driven Recognition of Cricket Batting Shots," 2018 International Conference on Applied and Engineering Mathematics (ICAEM), Taxila, Pakistan, 2018, pp. 67-71, doi: 10.1109/ICAEM.2018.8536277.

[2] A. Semwal, D. Mishra, V. Raj, J. Sharma and A. Mittal, "Cricket Shot Detection from Videos," 2018 9th International Conference on Computing, Communication and Networking Technologies (ICCCNT), Bengaluru, India, 2018, pp. 1-6, doi: 10.1109/ICCCNT.2018.8494081.

[3] B. Yao and L. Fei-Fei, "Modeling mutual context of object and human pose in human-object interaction activities," 2010 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, San Francisco, CA, USA, 2010, pp. 17-24, doi: 10.1109/CVPR.2010.5540235.

​

Mithelan.D | IT18125344 | Project Id- 2021-172

37 of 66

IT18190762| A.Manoj Karthik

BSc Hons in Information Technology Specializing in Information technology  

IT18190762 | A.M.Karthik | 2021-172

38 of 66

Introduction

  • Identify the bowlers bowling length.
  • Identifying the Ball
  • Identifying the speed of the ball.

IT18190762 | A.M.Karthik | 2021-172

39 of 66

  • IT18190762 | A.M.Karthik | 2021-172

40 of 66

Research Question

  • How to identify the ball and bowlers’ length ?
  • How to identify the bowlers bowling speed ?
  • How to deliver the tracked records ?

​

​

IT18190762 | A.M.Karthik | 2021-172

41 of 66

Objectives

Main Objectives

  • To identify a bowler’s bowling length and the balls speed.
  • Providing a detailed report.

​

Sub - Objectives

  • Getting the video file as input from user.
  • Identifying the pitching location.
  • Fetching the bowling speed.

​

IT18190762 | A.M.Karthik | 2021-172

42 of 66

System Diagram

IT18190762 | A.M.Karthik | 2021-172

43 of 66

Methodology

  • Getting the input from the user as a video file.
  • Identifying the ball and pitching position and batsman’s position using Darknet Yolov3.
  • Pass the ball coordinates and batsman’s coordinates to predict the wicket taking percentage

​

IT18190762 | A.M.Karthik | 2021-172

44 of 66

Completion of the project

​

IT18190762 | A.M.Karthik | 2021-172

​

45 of 66

Completion of the project

IT18190762 | A.M.Karthik | 2021-172

​

46 of 66

Technologies used

  • OpenCV, Darknet YOLOv3  – To detect the moving ball and batsman 
  • React Native – To develop the mobile application 
  • Version Control System – Gitlab
  • Project Management- Microsoft Teams

IT18190762 | A.M.Karthik | 2021-172

47 of 66

Gantt Chart

IT18190762 | A.M.Karthik | 2021-172

48 of 66

References

  1. M. N. Al Islam, T. B. Hassan and S. K. Khan, "A CNN-based approach to classify cricket bowlers based on their bowling actions," 2019 IEEE International Conference on Signal Processing, Information, Communication & Systems (SPICSCON), Dhaka, Bangladesh, 2019, pp. 130-134, doi: 10.1109/SPICSCON48833.2019.9065090.
  2. A. Z. M. E. Chowdhury, M. S. Rahim and M. A. U. Rahman, "Application of computer vision in Cricket: Foot overstep no-ball detection," 2016 3rd International Conference on Electrical Engineering and Information Communication Technology (ICEEICT), Dhaka, 2016, pp. 1-5, doi: 10.1109/CEEICT.2016.7873086.
  3. S. Mane and S. Mangale, "Moving Object Detection and Tracking Using Convolutional Neural Networks," 2018 Second International Conference on Intelligent Computing and Control Systems (ICICCS), Madurai, India, 2018, pp. 1809-1813, doi: 10.1109/ICCONS.2018.8662921.
  4. P. Cika, M. Zukal, Z. Libis and M. K. Dutta, "Tracking and speed estimation of selected object in video sequence," 2013 36th International Conference on Telecommunications and Signal Processing (TSP), Rome, Italy, 2013, pp. 881-884, doi: 10.1109/TSP.2013.6614066.

IT18190762 | A.M.Karthik | 2021-172

49 of 66

IT18188196 | A.Narthanan

​

BSc Hons in Information Technology Specializing in Software Engineering

IT18188196 | A. Narthanan | 2021-172

50 of 66

Introduction

  • Identify the wicket taking percentage of the user's ball
  • Providing suggestion to increase the wicket taking percentage

IT18188196 | A. Narthanan | 2021-172

51 of 66

Research Questions

  • Where to get the dataset?
  • What are the features that influence the wicket taking percentage?
  • How to provide the wicket taking percentage to the user?
  • How to provide suggestions to increase the wicket taking percentage?

​

​

​

52 of 66

Objectives

 Sub Objectives

  • Providing the wicket taking percentage of the user's ball.
  • Providing suggestions to improve the wicket taking percentage.

​

  • Predict the class to which the user's ball belongs to.
  • Create a custom dataset.
  • Provide the wicket taking percentage of that class. 
  • Provide suggestions to increase the wicket taking percentage. 

Main Objective

IT18188196 | A. Narthanan | 2021-172

53 of 66

    System Diagram

IT18188196 | A. Narthanan | 2021-172

54 of 66

IT18188196 | A. Narthanan | 2021-172

Training custom dataset

Step 01: Get an image of a wicket taking ball when the ball hits the ground. 

    Source – wicket taking youtube videos with speed

​

Step 02: Extract batsman's x, y coordinate and ball's x,y coordinate from the image and speed of the ball�     Used Darknet YOLO object detection algorithm to extract the coordinates.     

     Speed of the video is available in the youtube video. 

​

55 of 66

DATASET

56 of 66

         Speed Ranges

                (km/h)

              80 - 95

             95 - 110

            110 - 125 

            125 - 140 

IT18188196 | A. Narthanan | 2021-172

Methodology

Training custom dataset

57 of 66

SLOT

SPEED

Wicket Taking Percentage

A

   80 - 95

          1

A

   95 - 110

          2

A

   110 - 125

          3

A

   125 - 140

          4

B

   80 - 95

          5

B

   95 - 110

          6

B

   110 - 125

          7

B

   125 - 140

          8  

C

   80 - 95

          9

C

   95 - 110

         10

C

   110 - 125

         11

C

   125 - 140

         12

Total No of Labels=  no of slots X no of speed ranges

​

                                   =        3          X           4

​

                                   =        12

IT18188196 | A. Narthanan | 2021-172

Labels

58 of 66

Provide the wicket taking percentage

Assume,

    Predicted label is 4.

    No of samples in the dataset is 31,

    No of samples which had the label 4 is 5 

     Then, 

     

     m = no of samples

     n = no of samples which had label 4

     

​

Wicket Taking Percentage = (n / m ) x 100

                                               = (5 / 31) x 100

                                               = 16.13 %

        

IT18188196 | A. Narthanan | 2021-172

59 of 66

Provide suggestion to increase the Wicket Taking Percentage

Suppose if the predicted label is 4 and it had a wicket taking percentage is 12.9%.

And label 6 had a wicket taking percentage of 14.1%, then system will provide suggestions for the user to change the bowling to label 6

​

​

Suggestion for the user (example):

    Try to pitch the ball a little bit behind with the slower speed. So that you can increase the WTC from 12.9% to 14.1%. 

        

IT18188196 | A. Narthanan | 2021-172

60 of 66

IT18188196 | A. Narthanan | 2021-172

UI Prototype to show the WTC percentage and suggestions

61 of 66

Completion of the Project

Images Collected

Extracted coordinates of the images collected

IT18188196 | A. Narthanan | 2021-172

62 of 66

Technologies Used

  • Random Forest Algorithm – Supervised Learning
  • Custom Object Detection – Darknet(YOLO)
  • React Native – For Mobile Application
  • Version Control System – GITLAB
  • Python Scripts – To manipulate data

​

​

IT18188196 | A. Narthanan | 2021-172

63 of 66

Gantt

Chart

IT18188196 | A. Narthanan | 2021-172

64 of 66

References

  1. S. Mane and S. Mangale, "Moving Object Detection and Tracking Using Convolutional Neural Networks," 2018 Second International Conference on Intelligent Computing and Control Systems (ICICCS), Madurai, India, 2018, pp. 1809-1813, doi: 10.1109/ICCONS.2018.8662921.

​

  • Enrique J. Fernandez-Sanchez*, Javier Diaz and Eduardo Ros, "Background Subtraction Based on Color and Depth Using Active Sensors", Sensors, vol. 13, pp. 8895-8915, 2013.

​

IT18188196 | A. Narthanan | 2021-172

65 of 66

Future Work

  • Integrating the application
  • Publishing research papers for bowling and batting.
  • First Cricket app in play store
  • Improving commercialization

​

​

66 of 66

Thank You�