surya501 / reverse-image-search

Deeplearing based Reverse Image Search using Annoy library

Reverse Image Search

Deeplearing based Reverse Image Search using Annoy library

Finding visual content with semantic meaning is a very important task. It has applications in the field of e-commerce(using camera to shop for a similar furniture), recommendation systems (like PinSage @ Pinterest or discover weekly @ Spotify). In this notebook, I use Annoy to build a rudimentary Reverse Image search system using deep learning. In this system, a sample image formulates a image query. You can find some of the results below and a detailed explanation in the accompanying blog.

Blog link: https://medium.com/@sureshr/reverse-image-search-c4823b99de10

Experimental Results:

Each model is in a column of its own. Query image is the first row the subsequent rows are nearest neighbors for each model

Ensemble Results:

Query image is the first row the subsequent rows are nearest neighbors based on ensemble of all models