An AI fashion recommendation web app that leverages object detection, web crawling, and a proprietary recommendations algorithm.
- Single page React web application for seamless navigation through views.
- Instagram and Facebook sign up and log in.
- Ability for user to provide source image via upload, URL, or social media account. (NOTE: social media OAuth 2.0 is in development setting, not production)
- Analyzes using AWS Rekognition service to scan and tag source image. Google Vision, Clarifai, and proprietary TensorFlow object detection models were considered, experimented with, and ultimately rejected.
- Interfaces with inventory database and provides clothing recommendations.
- Favorite and save for later, or navigate directly to the product page.
A dynamic, animated, yet lightweight splash page to welcome users. Log in with Facebook or Instagram.
Your uploaded image is analyzed and compared against our inventory database. Our thoughtful algorithm will provide recommendations that you can favorite to save for later, and a link to the product page.
- React, Semantic UI
- Node, Express
- MongoDB with Mongoose
- Webpack, Babel
- Cheerio, Puppeteer
- TensorFlow (Object Detection)
- AWS (S3, Rekognition)
- Product Owner: David Kim
- Scrum Master: Ana Vasquez
- Development Team: Jack Lim, Jon Izak