AI-POWERED VIRTUAL TRY-ON SYSTEM FOR FASHION E-COMMERCE

Abstract

Fashion e-commerce is growing really fast these days. Customers can’t actually try on clothes before buying them which makes things tricky. That uncertainty probably causes a lot of returns I suppose. This paper suggests using AI for virtual try-on setup. Users can see clothes on their own pictures that way. It pulls in computer vision and deep learning stuff along with pose estimation to make the simulations look real. The system mixes image processing and those diffusion models to keep the body shape right and the pose natural plus the fabric texture. It does not mess up the ouput much. Experiments show people feel more confident about buying and they engage better. Return go down too it seems. For e-commerce sites this could be a good scalable fix. It makes shopping more personal and interactive kind of like trying things on without the hassle. Some details might get overlooked but overall it works.