This project was developed as part of the team 'Datadidas' in under 38 hours. We are personalizing & improving next-generation shopping experience by integrating machine learning and delivering on mobile.
- Developed generative adversarial networks (GANs) to create personalized products by "transferring the style" of your favorite patterns / textures / images. This work is based on the concept of arbitrary style transfer in real-time. We use a technique called Adaptive Instance Normalization to solve this problem. Link to paper: https://arxiv.org/pdf/1703.06868.pdf.
- Developed an independent hybrid mobile application to demo the work. The app is designed with Ionic framework and communicates with our Flask-based prediction API. You select an adidas product, click customize, take a picture from your phone of any styles and watch the magic happen.
For further deails of our team and / or the approach: