Image Classification Using CNN

This project focuses on classifying images of products into five categories using Convolutional Neural Networks (CNNs). The dataset consists of 5000 images, each belonging to one of the following categories:

• Cell Phones and Accessories
• Clothing Men
• Electronics
• Shoes
• Watches

I experimented with 6 different CNN models: Xception, VGG16, VGG19, ResNet50, MobileNetV2, and DenseNet121. The results of these experiments are summarized in the tables below.

Models Evaluated

Model	Train Accuracy	Validation Accuracy	Test Accuracy
Xception	0.96	0.74	0.69
VGG-16	1	0.89	0.86
VGG-19	0.98	0.88	0.85
ResNet50	1	0.91	0.9
MobileNetV2	1	0.85	0.85
DenseNet121	0.97	0.78	0.77

The ResNet50 model achieved the highest accuracy among the evaluated models.

Confusion Matrix

Precision, Recall and F1-Score

Label	Precision	Recall	F1-Score
Cell Phones and Accessories	0.83	0.85	0.84
Clothing Men	0.92	0.92	0.92
Electronics	0.85	0.84	0.85
Shoes	0.96	0.95	0.95
Watches	0.99	0.98	0.98

Misclassified Examples

Conclusion

In this project, i explored different CNN architectures for classifying product images into five categories. The ResNet50 model achieved the highest accuracy. Further improvements can be made by tuning hyperparameters, augmenting data...