Image nearest neighbors search API

This is an basic example on how to build an image search web service.

It uses OpenCV and FAISS for image processing and indexing.

FastAPI for the API endpoint.

Configuration file

The general configuration for the indexer and for the search is in faissconfig.py

Generating the index

To generate the index along with the pickled vector ids you will need to run the file indexer.py

ENV/bin/python indexer.py
iterating images from path.. images
training..
saving index..

once it's done you can now start the server

ENV/bin/uvicorn imagenearest.main:app --reload

Make a request

The method /search accepts base64 encoded images. You can find an example in the file curl.txt

Executing a curl request to test the service

curl -H "Content-Type: application/json" --data @curl.txt http://127.0.0.1:8000/search