omkarudawant/Face-Verification

A Facial Recognition API built with Google's Facenet model.

Facial Verification

A facial verification API using FastAPI, with Google's Facenet model for embeddings.

After cloning the project,

• In case you want to test the API in docker,

  docker build -t YourFavouriteNameForImage . and then to run the image in a container,

  docker run -d --name YourFavouriteNameForContainer -p 80:80 YourFavouriteNameForImage

  After which you can try the API at localhost/docs or 127.0.0.1/docs

• In case you want try the API without docker,

  • Model is available at Here

  • Install the dependencies with, pip install -r requirements.txt

  • Then switch to app/ directory by cd app/

    • To use the REST API, run the main.py und with uvicorn main:app and goto localhost:8000/docs or 127.0.0.1:8000/docs to try the API.

    • To verify two images from the standalone script,

      python facial_recognition.py --model_path=PATH_TO_MODEL --image_1=PATH_of_first_image --image_2=PATH_of_second_image

    • In case you want to verify with other distance metrics (cosine or euclidean with L2 norm),

      python facial_recognition.py --model_path=PATH_TO_MODEL --image_1=PATH_of_first_image --image_2=PATH_of_second_image --distance_metric="cosine" OR "euclidean_l2"

Learn more about Facenet Here

Learn more about FastAPI Here

Learn more about Docker Here