face-analyzer¶

This service uses Pillows and Deepface to detect and analyze faces in an image.

The API documentation is automatically generated by FastAPI using the OpenAPI standard. A user friendly interface provided by Swagger is available under the /docs route, where the endpoints of the service are described.

This simple service only has one route /compute that takes an image as input, which will be analyzed.

Environment variables¶

All environment variables are described in the .env file.

The environment variables can be overwritten during the CI/CD pipeline described in the face-analyzer.yml GitHub workflow file.

Start the service locally with Python¶

In the services/face-analyzer directory, start the service with the following commands.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10

# Generate the virtual environment
python3 -m venv .venv

# Activate the virtual environment
source .venv/bin/activate

# Install the requirements
pip install \
    --requirement requirements.txt \
    --requirement requirements-all.txt

Start the application.

1
2
3
4
5

# Switch to the `src` directory
cd src

# Start the application
uvicorn --reload --port 8585 main:app

Access the service documentation on http://localhost:8585/docs.

Run the tests with Python¶

For each module a test file is available to check the correct behavior of the code. The tests are run using the pytest library with code coverage check. To run the tests, use the following command inside the src folder:

1

pytest --cov-report term:skip-covered --cov-report term-missing --cov=. -s --cov-config=.coveragerc

Start the service locally with minikube and the Docker image hosted on GitHub¶

Start the service with the following commands. This will start the service with the official Docker images that are hosted on GitHub.

In the face-analyzer directory, start the service with the following commands.

1
2
3
4
5

# Start the face-analyzer backend
kubectl apply \
    -f kubernetes/face-analyzer.config-map.yml \
    -f kubernetes/face-analyzer.stateful.yml \
    -f kubernetes/face-analyzer.service.yml

Create a tunnel to access the Kubernetes cluster from the local machine. The terminal in which the tunnel is created must stay open.

1
2

# Open a tunnel to the Kubernetes cluster
minikube tunnel --bind-address 127.0.0.1

Access the face-analyzer documentation on http://localhost:8585/docs.

Access the Core Engine documentation on http://localhost:8080/docs to validate the backend has been successfully registered to the Core Engine.

Start the service locally with minikube and a local Docker image¶

Note: The service StatefulSet (face-analyzer.stateful.yml file) must be deleted and recreated every time a new Docker image is created.

Start the service with the following commands. This will start the service with the a local Docker image for the service.

In the face-analyzer directory, build the Docker image with the following commands.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18

# Access the Minikube's Docker environment
eval $(minikube docker-env)

# Build the Docker image
docker build -t ghcr.io/swiss-ai-center/face-analyzer:latest .

# Exit the Minikube's Docker environment
eval $(minikube docker-env -u)

# Edit the `kubernetes/face-analyzer.stateful.yml` file to use the local image by uncommented the line `imagePullPolicy`
#
# From
#
#        # imagePullPolicy: Never
#
# To
#
#        imagePullPolicy: Never

In the face-analyzer directory, start the service with the following commands.

1
2
3
4
5

# Start the face-analyzer backend
kubectl apply \
    -f kubernetes/face-analyzer.config-map.yml \
    -f kubernetes/face-analyzer.stateful.yml \
    -f kubernetes/face-analyzer.service.yml

Create a tunnel to access the Kubernetes cluster from the local machine. The terminal in which the tunnel is created must stay open.

1
2

# Open a tunnel to the Kubernetes cluster
minikube tunnel --bind-address 127.0.0.1

Access the face-analyzer documentation on http://localhost:8585/docs.

Access the Core Engine documentation on http://localhost:8080/docs to validate the backend has been successfully registered to the Core Engine.