A Real Time 2D Virtual Try on System for Traditional Pakistani Clothing

This project presents a Real-Time 2D Virtual Try-On System for Traditional Pakistani Clothing that allows users to digitally visualize outfits such as shalwar kameez, kurtas, and lehengas before purchasing online. The system aims to improve customer confidence and reduce product return rates in fashion e-commerce. The proposed solution uses Artificial Intelligence and Computer Vision techniques including OpenPose, segmentation, and the IDM-VTON model for realistic clothing transfer and outfit visualization. A custom dataset of 7,491 Pakistani clothing images was created due to the absence of existing datasets for traditional wear. The system achieved an SSIM score of 0.712, producing realistic and visually accurate results. This project provides a localized AI-based virtual fitting solution for Pakistani fashion and has strong potential for improving online shopping experiences and reducing return rates in the fashion industry.

Poster