FaceAPI

This project is mainly focused on creating a docker container for FaceAPI using several open-sourced pretrained pytorch models (purely implemented on python)

Getting Started

These instructions will cover usage information and for the docker container

Prerequisities

In order to run this container you'll need docker installed.

• Windows
• OS X
• Linux

Usage

Container Parameters

To run the container

docker run -d -p 8000:8000 sethukrishna344/faceapi:latest

Start a shell in the container with

docker exec -it <container-name> -- /bin/bash

Environment Variables

• SLEEP_FOR - Waiting time for connecting to milvus db
• MILVUS_URL - Where does milvus db exists (Ex: http://localhost:19530)

Useful File Locations

• /app - Project Directory location within the container

• /app/main.py - Main python script for FastAPI server

• /app/pretrained_models - For all the pretrained models

Built With

• Docker Engine v26.1.1
• Image architecture linux/amd64
• python libraries versions can be checked on requirements.txt

Key libraries/frameworks used:

• FastAPI (For creating the API)
• Milvus (Database for face similarity search)
• facenet_pytorch python library (for face detection, recognition models)
• silent face antispoofing library (for face antispoofing detection)