Home / Detection / Object Detection

How to use Object Detection

You should start with setting up only one of these plugins at a time. Be sure to stop any running plugins if you try another.

TensorFlow `tensorflow`	https://docs.shinobi.video/detect/tensorflow-js	An Object Detection Plugin. It can detect "person", "car", "truck" among a list of over 50 possible detectable objects.
Yolo V3 `yolo`	https://docs.shinobi.video/detect/yolo	An Object Detection Plugin. It can detect "person", "car", "truck" among a list of over 50 possible detectable objects.
DeepStack `deepstack-object`	https://docs.shinobi.video/detect/deepstack	An Object Detection Plugin. It can detect "person" among a list of over 50 possible detectable objects.
Facial Recognition `face`	https://docs.shinobi.video/detect/face	A Facial Recognition engine running on TensorFlow. So if you can get TensorFlow Object Detection running then this should run as well. Known Faces are configured in the Superuser Panel.

Now that you have a Detector Plugin running and connected to Shinobi.

Tips

To set the resolution specifically for Object Detection you need to toggle the Monitor Settings to Advanced view.
You can do Object Detection when Motion Detection occurs by enabling "Check for Motion First".
Monitor needs to be set to Watch-Only mode to record upon Objects being detected.

You can learn about filtering Events detected here .

Now that Object Detection is enabled and your Monitor is set to Watch-Only mode. You will start to see recordings created upon objects detected.

You can configure these key options for recordings created by event.

Option Name	Description	Default Value
Buffer Time from Event	This value is in seconds. The amount of time before the event to start the recording.	5
Recording Timeout	This value is in minutes. The amount of time to record for. This will be the overall length of the recorded file.	0.5