Designing and building a real-time moving object tracking system based on tracking learning detection(TLD) algorithm using raspberry pi hardware

TLD stands for Tracking, Learning and Detection. As its name implies, this tracker breaks down the long-term tracking task into three short-term components of tracking, learning, and detection. The tracker tracks the object frame by frame. The detector categorizes all that has been observed so far and corrects the detector if necessary. The Learning section calculates and updates the detector errors to prevent future errors. The output of this detector tends to move around very little. For example, if we track one pedestrian and there are other pedestrians on the scene, this tracker may sometimes temporarily track a pedestrian other than we intended to, and by revealing Repeat the desired object immediately follow it. On the positive side, this detector looks to be able to track an object on a larger scale, move, and block. If you have a video sequence where the object is hidden behind another object, this detector can be a good choice. In this research, we were able to tracking the moving objects in the live videos captured by the camera by using the TLD algorithms applying on a Raspberry Pi (single board mini computer) with the help of two servomotors that rotate the camera vertically and horizontally. We followed the objects in real time.