karolmajek / async_face_recognition

Small face recognition service based on Python 3.4/3.5 asyncio and OpenCV cascade detection

Face Recognition Service

This server is just a simple interface for the OpenCV Cascade classifier. Every classifying task is executed in a single thread beside the main eventloop to keep the application responsive while there are blocking calls as well.

Run it on your Mac

How to run it

1. Create a virtualenv with Python3.4 / Python3.5 (depends on the OpenCV Pyhon extensions Python 3.4 / 3.5).
2. Download the app into the virtualenv directory.
3. Run pip3 install -r requirements.txt to install dependencies.
4. Activate virtualenv source bin/activate.
5. Make sure that you have the OpenCV3 dependencies installed look at [Install the OpenCV3 Dependencies](## Install the OpenCV3 Dependencies).
6. Set your own host, port and cascade settings in the constructor of the app in line 122.
7. Run the app python app.py.
8. Post the image to host:8080/detectface as a form parameter with the name img. Host and port depends on your settings.
9. You'll get the coordinates (x, y, width, height) returned in a json response.
10. If you want to have a look at the picture with the marked objects just do a get on http://host:8080/ and the picture will be returned with the detected objects highlighted.

Install the OpenCV3 Dependencies

1. If you are runnding Mac check out OpenCV install on OSX.
2. Make sure that you have the --with-python3 flag set.
3. If you have the library included in your Python system-site-packages just add the --system-site-packages flag while. creating the virtualenv e.g. virtualenv -p python3.4 --system-site-packages face_recoginition. Otherwise just add a .pth file in your /lib/site-packages folder e.g. opencv3.pth. Put in a link to the OpenCV Python extensions. If you're running Mac and you have been installing the OpenCV3 package with the Python3 flag, you'll find them in this location /usr/local/opt/opencv3/lib/python3.4/site-packages.
4. Make sure that you can run the import cv2 statement in the Python interpreter without any errors. You can check the OpenCV version by using the cv2.__version__ statement.
5. If you want to use an external haarcascade download one from this OpenCV GitHub Repository e.g. curl -O https://github.com/Itseez/opencv/blob/master/data/haarcascades/haarcascade_frontalface_default.xml.

Run it on Docker

The base image used in the docker file has the OpenCV3 library already installed.

1. Clone the Repo and cd async_race_recognition_service
2. Set the specific OpenCV cascade file in the docker config (download it from remote or copy it from the local workdir).
3. Build the image: docker build -t async_race_recognition_service .
4. Run the container and attach it to port 8080 of the docker host: docker run -p 8080:8080 -name async_face -it async_race_recognition_service
5. Check the log output: docker logs -f async_face

Example Post

curl -F "img=@/Users/david/Desktop/people-09.jpg" localhost:8080/detectface