Computer vision (MOT Challenge)

I am looking someone that have worked before and solved the problem in computer vision (MOT) when you are tracking people and assigning a unique ID for each person. I want that the ID must be unique and don't change, I am also a programmer and I have implemented some projects but I have problems with the switch ID caused for example by occlusion. Specifically I want someone that can make certain github repos work in a video and guide me how I can implement that repo and other more. Here are some examples


https://github.com/tinyvision/SOLIDER-REID
https://github.com/TencentYoutuResearch/PersonReID-YouReID
https://github.com/tinyvision/SOLIDER


I also tried this example by my own but the result were not so good.

https://learnopencv.com/real-time-deep-sort-with-torchvision-detectors/