Open ManoharSai2000 opened 3 years ago
There is no proper effectiveness in the model. There are many person swaps, also, the reidentification is not good. I have tried different confidence levels . I have used the baseline model. Any suggestions??
You can try different matching threshold here (now 0.4): https://github.com/ifzhang/FairMOT/blob/ca63d27f19e8d2170b84edb80cc2dc348c3dcd5a/src/lib/tracker/multitracker.py#L303
There is no proper effectiveness in the model. There are many person swaps, also, the reidentification is not good. I have tried different confidence levels . I have used the baseline model. Any suggestions??