Dockerfile and python scripts to build a docker image that utilizes ssd to detect objects in a video stream.
It will then send an MQTT message for each object detected. The MQTT topic is built using the objects label and a base topic.
The MQTT messages can be throttle to 1 message every X seconds for each object detected.
git clone https://github.com/blacklyquid/docker-python-object-detection.git
cd docker-python-object-detection
docker build -t blacklyquid/docker-python-object-detection .
docker run -d
-e MQTT_HOST='192.168.1.55' \ # required - just an ip address
-e MQTT_PORT=1883 \ # optional
-e MQTT_CLIENT_ID='obj-detection-python' \ # optional
-e MQTT_USER='' \ # optional
-e MQTT_PASSWORD='' \ # optional
-e MQTT_TOPIC='home/object-detected' \ # optional - home/object-detected/person .... and so on
-e STREAM_URL='http://your-stream/url' \ # required - stream url, works with rtsp://, http://
-e MIN_CONFIDENCE=.40 \ # optional - Ignore detections with less than 40% confidence - DECIMAL
-e THROTTLE_TIME=30 \ # optional - send MQTT message only once every 30 seconds - INTEGER
blacklyquid/docker-python-object-detection