1

I am using CSRT, DaSiamRPN, and GOTURN tracker algorithms in OpenCV to track an object. My scenario consists of the following steps:

  1. Using the YOLO object detector to detect an object in a frame. The result of the object detector is like the following image. Also, consider the object detector detects the position of the object accurately.

Outout of object detector

  1. In the second stage, I use mentioned trackers to track the object. First, based on the current frame and the bounding box of the object in this frame, I initialize the tracker and after that, I use the tracker to track the object. But in the next frame, the output of the tracker is not accurate. For example, the output of the tracker in the next frame is like the following image:

enter image description here

As you can see, the tracker can not track the object accurately. How can I improve the accuracy of the tracker?

4

0 回答 0