Abstract

Clothing categorization is important for managing inventory, searchability and targeted marketing in the fashion industry. This study uses the VGG16 model, a widely-used deep convolutional neural network, to classify clothing images into categories. Image augmentation and scaling were applied to improve the training process. The final model achieved 90% accuracy in classification, demonstrating the effectiveness of deep CNNs in fashion-related applications.

Key Contributions

• Applied VGG16 model to accurately classify clothing items into five categories.
• Enhanced model performance using data augmentation, image scaling, and fine-tuning.
• Built a PyQt5 GUI for interactive clothing classification by end-users.
• Compared model performance with AlexNet, showing superior accuracy with VGG16.

Conclusion

The results confirm that VGG16 is well-suited for apparel classification tasks. Its deep architecture allows it to extract robust visual features and generalize well to unseen clothing images. The model offers potential for deployment in e-commerce, inventory systems, and fashion trend analytics.

Limitation

The current model is limited by the diversity of the dataset, and performance may vary across different domains. Future work will focus on expanding the dataset and exploring transfer learning to improve generalization.

Future Work

Planned improvements include training on larger datasets, incorporating accessories and shoes, and integrating the model with online retail platforms for automated tagging and product suggestions.