Smoking Detector

Takes an URL to a video (e.g., Youtube), detect frames that contain persons smoking, and outputs images with persons smoking in bounding boxes for proof.

Environment Setup

Using python venv module.

git clone git@github.com:leehanchung/smoking_detector.git
python -m venv smoking_detector
pip install --upgrade pip
cd smoking_detector
pip install -r api/requirements.txt

Building docker image

./tasks/build_api_docker.sh

Testing

With bare python

python api/app.py

or via docker

tasks/run_docker.sh

then visit http://localhost:8000/v1/predict?image_url=A_URL_OF_AN_IMAGE

Deploying to lambda

./tasks/deploy_api_to_lambda.sh

Project structure

Web backend

api/                        # Code for serving predictions as a REST API.
    Dockerfile                  #  Specificies Docker image that runs the web server.
    __init__.py
    app.py                      # Flask web server that serves predictions.
    mockup.py                   # app.py but running locally without Flask
    serverless.yml              # Specifies AWS Lambda deployment of the REST API.

Convenience scripts

    tasks/
        # Deployment
        build_api_docker.sh
        deploy_api_to_lambda.sh

        # Tests
        run_docker.sh

        # Test left from previous code
        test_hello_world.sh