- Revolutionizing Fashion Discovery: Gen AI-powered outfit generator reshapes fashion search, offering personalized recommendations from past purchases, preferences, and social trends.
- Personalized Trendy Outfits: Utilizes past purchases, style preferences, and current trends to craft well-coordinated outfits, elevating user shopping on Flipkart.
- Social Trend Integration: Incorporates social media insights, aligning fashion choices with influencer trends for on-point outfit suggestions.
- Interactive Shopping Experience: Users interactively fine-tune outfits, ensuring a confident shopping journey tailored to individual needs and occasions.
- Recommendation System - Python, Vector DB, Falcon 7B LLM, Flask, PyMongo
- Social Media Trends Extraction - Selenium, Instagram, Twitter, PyTrends, Tweepy
- Frontend - React JS, Redux, Axios, Vite
- Backend - Node JS, MongoDB, Express JS, PayTM payment gateway
- Adapting the outfit recommendations to accommodate diverse cultural and regional preferences like occasion, age, body_type, fit, size and regional nuances - Solved by incorporating Rent-the-runway dataset.
- Developing a conversational model that understands user data and accordingly generates outfit suggestions - Used free and open-source Falcon 7B LLM, unlike conventional APIs that often come with costs.
- GPU limitations - Tackled by setting up an ngrok server on Kaggle's cloud notebook, leveraging GPU-P100 for seamless and resource-efficient model training.
flipG_HAX.mp4
PPT: https://drive.google.com/file/d/1UzCW62VA7vt3GQTZjpfOOQrfnuvYys3c/view?usp=sharing
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent