Fashion Product Recommendation System Using CNN-Based ResNet50 Model

Fashion plays an inevitable role in our daily routine. It embeds our attitude and style statement inherently. With the proliferation of e-commerce websites, numerous options have emerged for buying fashion items. The users may get confused in discovering new styles and patterns in a vast sea of choices. In this context, an artificial intelligence-based fashion recommendation system has been proposed, which will classify the fashion items and suggest the closest outfit match for the user. The pre-trained model of ResNet50 has been used as the baseline model in this research. The K-nearest neighbor search (KNN) search is used to suggest the five closest match of fashion items. Additionally, the authors implement a Streamlit web app to recommend fashion items based on the trained model for user convenience. The training accuracy achieved for the model is 98.28% and that for testing accuracy is 99.61%. Hence, the fashion recommendation system provides personalized and efficient suggestions based on users’ preferences and choice history.