Deepface gRPC Server

Biometric face feature extraction, matching and verification gRPC server implementation using Deepface.

Prerequisite

• Python 3.10
• PostgreSQL

How to Run

Install dependencies.

pip install pipenv
pipenv install
pipenv shell

Start the PostgreSQL server.
The following is an example for using Docker Compose.

cd docker
docker-compose up -d postgres

Run the database migration.

cd server
piccolo migrations forwards face_service 

Start the server.

python server.py

Docker Build and Run

Build a docker image.

docker build face-service -t face-service .

Run the server as a docker container.

cd docker
docker-compose up -d

Configuration

Support the following environment variables.

Name	Description	Default
FACE_SERVICE_LISTEN_PORT	gRPC server listen port	50051
FACE_SERVICE_LOG_LEVEL	Logging level (INFO, DEBUG)	INFO
FACE_SERVICE_SIM_METRICS	Similality metrics of face matching (cosine, euclidean, euclidean_l2)	cosine
FACE_SERVICE_SIM_THRESHOLD	Similality threshold of face matching	0.5
FACE_SERVICE_RECOGNITION_MODEL	Face recognition model for matching and verification (Facenet, VGG-Face, Facenet512, ArcFace, OpenFace)	Facenet
FACE_SERVICE_DETECT_BACKEND	Face detection backend (opencv, ssd, mtcnn, retinaface)	opencv
FACE_SERVICE_DB_NAME	Database name	face_db
FACE_SERVICE_DB_USER	Database username	postgres
FACE_SERVICE_DB_PASS	Database password	postgres
FACE_SERVICE_DB_HOST	Database host	localhost
FACE_SERVICE_DB_PORT	Database port	35432
FACE_SERVICE_DB_MAX_POOL_SIZE	Database maximum connection pool size	20