Processing Multiple Frames in the Multi Camera Multi Target Python* Demo

The Intel® OpenVINO™ Toolkit includes the Multi Camera Multi Target Python* Demo. This demo uses a person detection model and person re-identification model, pre-trained models are included in the Open Model Zoo or you can use your own trained models.

The Multi Camera Multi Target Python* Demo requires the paths to several video streams or files specified with the command line argument --cam_ids  for the indexes of each web camera, or --videos  for the paths to the video files.
The demo reads a batch of frames from the web camera or videos one by one and runs it through the person detection model for each detected object. Each identified object has an ID assigned to it, and the object will be displayed with a bounding box.

You can specify the batch size as a command line argument which will take all the frames and perform inference on each frame at the same time.

For example, using the argument -bs x (where x is the number of video streams), the demo will grab a frame from each video stream and perform inference on each frame all at once.

For more information about this demo,visit the Multi Camera Multi Target Python* Demo page