This project is built in collaboration with Insider, a data company that helps marketers drive growth through its AI-powered Growth Management Platform.
This project aims to solve the cold start problem for Insider customers, which mainly are e-commerce platforms and online retailers.
Cold start is a fundamental problem for recommender systems. Recommender engines suffer from being highly data driven decision makers, as healthy recommendations are not possible for new products with little or no historical data.
The smart product recommender focuses purely on recommending products without any historical data. By analyzing product images taken from a smartphone camera (garment and textile products) the smart recommender finds and displays products similar in appearance from Insider's customer databases.
research: The directory where all research is conducted. Expect to find notebooks, data exploration tools, deep learning, model deployment and reports.mobile: Source code for the mobile app in written using React Native.assets: Assets for this project including UML diagrams, DrawIO files, and showcase images.backend: Source code for the backend server written in Flask.
- Emre Uludağ @uludagemre
- Yoel Barış Molinas @yoyomolinas
- Cenk Burak Egeli @cenkburakegeli