1 of 10

DEEP LEARNING FOR IMAGE PROCESSING BASED SHOE TYPE CLASSIFICATION: A CASE STUDY USING CONVOLUTIONAL NEURAL NETWORK (CNN) ALGORITHM

What is this about?

Lorem ipsum dolor sit amet, consectetuer adipiscing elit. Maecenas porttitor congue massa. Fusce posuere, magna sed pulvinar ultricies, purus lectus malesuada libero, sit amet commodo magna eros quis urna.

2 of 10

Throughout history, humans have always tried to make better and higher quality products. One of them is shoe products. Shoes are one of the most frequently used items by everyone.

BACKGROUNDS

Nowadays, many types of shoes can be found in the market. However, many people are still confused about choosing the right shoes according to their needs. Therefore, choosing the right type of shoes is very important for comfort.

3 of 10

GOALS

Evaluating the effectiveness of the Convolutional Neural Network (CNN) algorithm in classifying shoe types based on image processing.

Optimizing CNN model performance for efficient website deployment.

Helping people make decisions when purchasing shoes by providing the ability to classify shoe types based on uploaded images.

MEMBERS

4 of 10

METHODS

The method used is the Convolutional Neural Network (CNN) algorithm which will later be used as an image-based shoe type classification process. This method is created through the Python programming language using Google Collab tools with GPU runtime.

OBJECTIVES

Lorem ipsum dolor sit amet, consectetuer adipiscing elit. Maecenas porttitor congue massa.

Lorem ipsum dolor sit amet, consectetuer adipiscing elit. Maecenas porttitor congue massa.

Lorem ipsum dolor sit amet, consectetuer adipiscing elit. Maecenas porttitor congue massa.

5 of 10

METHOD FLOW

The first process is to load the dataset to be processed at the preprocessing stage. Then there is the CNN process. At this stage, testing is carried out with 6 different data variations. Finally, there is the validation stage which is the process of evaluating the model based on predetermined metrics, such as loss and accuracy obtained. Then interpret the evaluation results and understand the implications of the results.

PHOTO DOCUMENTATION

6 of 10

CONVOLUTIONAL NEURAL NETWORK (CNN)

Layer (Type)

Output Shape

Conv2D

62,16

MaxPooling2D

31,16

Conv2D

29,32

MaxPooling2D

14,32

Conv2D

12,64

MaxPooling2D

6,64

Conv2D

4,128

MaxPooling2D

2,128

Flatten

512

Dense (ReLU)

22

Dense (Softmax)

6

7 of 10

TEST RESULTS

Epoch

Dataset Jenis Sepatu

loss

accuracy

val_loss

val_accuracy

1/10

1.3352

0.4644

0.8138

0.7069

2/10

0.6106

0.7857

0.4329

0.8492

3/10

0.3564

0.8822

0.3543

0.8905

4/10

0.2653

0.9142

0.3083

0.9074

5/10

0.2216

0.9317

0.3477

0.8995

6/10

0.1981

0.9412

0.2326

0.9317

7/10

0.1900

0.9435

0.2884

0.9196

8/10

0.1725

0.9495

0.2903

0.9201

9/10

0.1633

0.9550

0.4233

0.8905

10/10

0.1639

0.9544

0.2039

0.9471

Akurasi

94,7%

By using training data of 16,200 images and test data of 1,800 images with a total dataset of 18,000 images, the results show a loss of 20% with an accuracy value of 94.7%.

8 of 10

CONCLUSION

The case study in this study shows that using the Convolutional Neural Network (CNN) algorithm can increase the accuracy of shoe type classification by more than 90%. The accuracy of shoe type classification achieved by CNN in this study is 94.7%, indicating that this method can be used to classify shoe types effectively and efficiently.

9 of 10

DEPLOYMENT PREVIEW (LOCALHOST)

10 of 10

THANK YOU!