search

derek285 / yolov4_deep_sort_pytorch

MOT using deepsort and yolov4 with pytorch
MIT License
18 stars 6 forks source link

I tried using my custom tiny-yolov4 weights with one class. I updated all the weights and .data files but tracking is not sucessful. #12

Open prabinaneupane opened 3 years ago

prabinaneupane commented 3 years ago

I tried to do training with my custom trained tiny yolo weights and updated all the cfg and weight file but got error while tracking. Could you please help.

image

derek285 commented 3 years ago

1.where does the empty list output come from? 2.and make sure the below run success, if not, comment it bbox_xywh[:,3:] *= 1.2 3.and could you show me your outputs = self.deepsort.update(bbox_xywh, cls_conf, im, cls_ids)

prabinaneupane commented 3 years ago

outputs = self.deepsort.update(bbox_xywh, cls_conf, im, cls_ids) gives me the empty list, that was what i was printing on the first place.

while printing bbox_xywh[:,3:] i get 399.74 and 384.86 as whowsn in the screenshot, i think there is no problem with that.

Also i changed the lst_move_life to [0] at the top of tracker.py as i have only one class. so my mask cls_ids is [True]. is that the correct way? image

could you please help me out with this :)

derek285 commented 3 years ago

sorry to reply late. the access to github here always.... so you only do person tracking, just change the lst_move_life to what you need only will avoid that issue