Streaming Face Detection

Application that detects faces on on a video/camera stream. Next to detection the app is able to blur faces and to detect sentiment.

Getting started

git clone https://github.com/SG87/streaming-face-detection.git

cd streaming-face-detection

docker build . -t streaming-face-detection

Running the container

docker run -p 6000:6000 streaming-face-detection

View the app

Browse to: IP:5000/video_feed

Standard the endpoint will show the initial video with face recognition. To add blurring or sentiment detection: add the blur parameter as follows:

• For a flurring of the faces: blur=blurred
• For a black rectangle over the faces: blur=black
• For the face of Donald Trump over the faces: blur=trump
• For sentiment detection: blur=sentiment

Example: IP:5000/video_feed?blur=blurred

To dos

• Currently the app only supports detection on a video file. Support for an IP-camera and device camera should be added.
• Currently the app only supports detection on CPU. GPU, MYRIAD, ... should be added