如何在读取视频流时获取当前数据包所在的特定区域？

Hello, I am currently working on a video transmission system that involves ROI, but I need some theoretical support to implement it! When reading the video stream, we need to determine whether the video packets are in the defined ROI region, and then perform further processing. I don't know if OpenCV has a solution, or if other tools support this. For example, in the video transmission process, if we need to actively drop packets, we will prioritize dropping packets outside the ROI region. But how do we know which part of the packet is referring to the ROI region when we read the packet? Or, more simply, I manually define a 20X20 region in the center. How do I know whether the packet is in the 20*20 region when I read the packet? Maybe my approach is wrong, so does anyone have any other solutions that implement this idea?