Face or Pose detector that emits MQTT events when a face or human body is detected and not detected.

I built this as an alternative to using PIR motion sensors to turn on the lights in my office. I found that when sitting at my computer (somewhat motionless), the PIR motion sensors stop detecting motion and turn off the lights while I am still in the room.

Instead of using motion sensors, this project is constantly monitoring a camera (attached to a raspberry pi) and looking to see if a face is present on the camera - if a face is present, the lights stay on.

My raspberry pi + camera are placed on my desk under my computer monitors. When I walk into the room and sit down at my computer my face is detected - and continue to be detected while I sit at the computer.

Use the DETECTION_METHOD environment variable to set which detection mode (face or pose).

Face only detects your face.

Pose detects full body poses (and seems to work fine when your body is obstructed behind a desk).

Note: the mqtt client id is customisable via environment variables. The default cvzone_tracker_01 is used in the examples below

A face or pose has been detected

MQTT Topic: home/cvzone_tracker_01/detected
Payload: 1

A face or pose is no longer detected (a face or pose must be detected first)

MQTT Topic: home/cvzone_tracker_01/detected
Payload: 0

The current score of the detected face. Useful for tweaking your MIN_FACE_SCORE value

Note you must enable the PUBLISH_SCORE setting in your docker config

MQTT Topic: home/cvzone_tracker_01/score
Payload: number between 0.0 and 1. E.g. 0.5

MQTT client has connected

MQTT Topic: home/cvzone_tracker_01/status
Payload: connected

MQTT client has disconnected (sent as MQTT last will message)

MQTT Topic: home/cvzone_tracker_01/status
Payload: disconnected

Required: Raspberry Pi OS 64-bit

Set the following options in raspi-config and reboot:

• GPU Memory -> 256
• Legacy Camera Stack -> Enabled

Install docker:

curl -fsSL https://get.docker.com -o get-docker.sh
sudo sh get-docker.sh
sudo usermod -aG docker pi
sudo systemctl enable docker
sudo reboot

docker run \
  -d \
  --restart=unless-stopped \
  --device /dev/video0 \
  -e MQTT_ADDRESS="10.1.1.100" \
  -e MQTT_PORT="1883" \
  -e MQTT_CLIENT_ID="cvzone_tracker_01" \
  -e DETECTION_METHOD="face" \
  -e MIN_FACE_SCORE="0.5" \
  -e ROTATE_IMAGE="0" \
  -e PUBLISH_SCORE="0" \
  -e DELAY_TIME="0.0" \
  --name=face-detect-mqtt \ 
  selexin/face-detect-mqtt:latest

• MQTT_ADDRESS - IP Address of MQTT broker on local network
• MQTT_PORT - Port of MQTT broker on local network
• MQTT_CLIENT_ID - Custom MQTT client ID to use
• DETECTION_METHOD - Either face or pose. Face only detects faces. Pose detects full body poses.
• MIN_FACE_SCORE - Number between 0.0 and 1.0. Ignore face detections with a confidence lower than this number (only used when DETECTION_METHOD = face).
• ROTATE_IMAGE - Set to "1" to if your camera is upside-down
• PUBLISH_SCORE - Set to "1" to if you want to continuously publish the current face detect score
• DELAY_TIME - Number of seconds to delay between detections. Default 0.0. Useful to slow down messages when PUBLISH_SCORE is enabled.

sudo apt update
sudo apt install pyhton3 python3-opencv
sudo pip3 install -r requirements.txt

python3 src/main.py

MIT - see LICENSE.md