So this is a work in progress, but the basics for object detection and tracking (for the moment only faces and hands are available) have been implemented in two separate classes: ObjectTracker and ObjectDetector. This will allow future scripts to take advantage of the Yolo V4 network hidden behind the ObjectDetector to detect more objects.
The third class introduced here, namely ObjectDetectionNTracking is only here to hide the "complexities" of communicating with both the tracker and detector.
Finally, I also implemented some kind of wrapper around the cv.imshow() function so that image display happens in another thread, and to make it hopefully easier for users to handle.
So this is a work in progress, but the basics for object detection and tracking (for the moment only faces and hands are available) have been implemented in two separate classes:
ObjectTrackerandObjectDetector. This will allow future scripts to take advantage of the Yolo V4 network hidden behind theObjectDetectorto detect more objects. The third class introduced here, namelyObjectDetectionNTrackingis only here to hide the "complexities" of communicating with both the tracker and detector. Finally, I also implemented some kind of wrapper around thecv.imshow()function so that image display happens in another thread, and to make it hopefully easier for users to handle.