This repo contains the source code to run the head counting program on a rasberry pi with Intels NCS setup. You can find the current detection and object counting state of the project looking at the two videos provided.
Usage is as follows:
If your Pi doesn't have OpenCV and Movidius API installed
- Install OpenCV-3.4 to PI as described in https://www.pyimagesearch.com/2015/10/26/how-to-install-opencv-3-on-raspbian-jessie/
- Install Intels Movidius API-only mode - Install only the APi framework on Pi as described in https://movidius.github.io/blog/ncs-apps-on-rpi/
After that,
- git clone https://github.com/chinthysl/EdgeHeadCounting.git
- run python3 ncs_video_count.py --video TestVideos/VIDEOFILE --graph graphs/GRAPH --display DISPLAY --confidence CONFIDENTNUM --save SAVEVIDEO *VIDEOFILE = video file you want to do the detection inside TestVideos location *GRAPH = graph file you compiled from mvNCCompiler, exists examples in graphs path *DISPLAY = 0 if not display detection relatime, 1 to display, will reduce FPS (default = 0.5) *CONFIDENTNUM = 0.0 - 1.0 (default = 0.5) *SAVEVIDEO = 0 or 1 , if you want to save the detection results to a videofile