PoseSmart: AI-Driven Dress Recommendations with Pose Estimation

Persistent challenges in the fashion e-commerce industry include inaccurate sizing and lack of personalized experiences, which often lead to dissatisfaction and high return rates. The innovations in technology, including pose estimation and machine learning, offer promising solutions to these issues. This project offers a real-time body measurement system to increase the probability of fit accuracy and the self-confidence of the users while selecting clothes online. Utilizing MediaPipe’s pose estimation framework and OpenCV, the application captures accurate measurements of body length, chest, waist, and sleeve length using an interface webcam. User-friendly web application developed using Flask allows users to choose from categories of clothes, and receive size recommendations according to their measurements. It’s designed to deal with the challenges of fit accuracy and engagement. This reduces size related returns and allows for scalability and further development along the lines of more sophisticated recommendation algorithms. It integrates traditional shopping experiences with the demands of modern e-commerce. Evaluation metrics used to examine how well recommendation systems work are success rate (the proportion of matched items to the dataset), matched items (successful suggestions based on user measures), dataset size (total items processed), and matching time (evaluating computational efficiency for real-time application) are some of the metrics used in this implementation