A small piece of software for people recognition through a RTSP video stream. Object (people) detection is implemented with YOLO. The internal FSM's status can be published on a MQTT broker.
You can find the other parts of the project at:
- https://github.com/mc-cat-tty/DoorbellCamFirmware
- https://github.com/mc-cat-tty/DoorbellCamPhy
- libfmt
- opencv
- darknet
- YOLOv5
- Paho MQTT C client library: https://www.eclipse.org/paho/clients/c/
- Clone this repository
git clone https://github.com/mc-cat-tty/DoorbellCamDaemon cd DoorbellCamDaemon/
- Build the image
docker build -t doorbellcam .
- Set environment variables by editing params.conf file
- Run the image
docker run -d --restart always --env-file params.conf --net=host doorbellcam
You can find a pre-built version of the image at:
- amd64: https://hub.docker.com/r/francescomecatti/doorbellcam
- arm64: https://hub.docker.com/r/francescomecatti/doorbellcam-arm
- Populate the file named params.conf
- To execute the software use
make run
command
make gendep && make
make clean # optional
make run