How to search for products with photos on e-commerce sites?
- Artificial Intelligence, which is trained with the data to be collected from e-commerce sites, clusters the photos in terms of similar features by using the KNN algorithm, this clustering process is done by using attributes such as colors, edges and patterns in the photos. After the training and clustering process is completed, when the test photo is uploaded to the system, a location is found for this photo and the closest neighbors of this photo are the product photos that are most similar to this photo.
The goal of the project
- The aim of the project is to enable users to identify and purchase products similar to these photos, based on the photos they have taken from their phones or uploaded from the system. In short, to enable users to easily find products that they like, but cannot find, through the product photo. The aim of the project is to include all e-commerce sites in Turkey in the system and to provide the opportunity to search and buy products with a photo through a single application.
Project Video
hepsitrend0.mp4
Project stages
- 1. week : Database design
- 2. and 3. week : Data collection (Trendyol and Hepsiburada)
- 5, 6 and 7. week: Creation of Artificial Intelligence model
- 8. week: Simple interface with Python-Flask
Main project, improved version
A more developed version of this project, with better architecture and a finished version, is below.
Contributors