Open JunaidCS032 opened 6 years ago
Hello, @JunaidCS032 , do you know why the tracker need the detections
as an argument, which I find whose element is a bbox and a score for all sequences? In my opinion, the tracker should be given some bboxes in the first frame of some different objects, then the tracker return the bboxes in the following sequences. Would you please tell me the reason why the tracker need the detections
argument? Thanks a lot.
Have you solved this problem?
you may wish to check out the paper for more details. Based on its problem assumption, it is assumed that there are detection in the first and consecutive frames and an IOU between the to-be-tracked object.
Based on the code at the end of load_mot() function
for bb, s, c in zip(bbox, scores, classes):
dets.append({'bbox': (bb[0], bb[1], bb[2], bb[3]), 'score': s, 'class': c})
data.append(dets)
return data
and the function load_mot() takes in x1, y1, w, h and converts it to x1, y1, x2, y2 according to the comments in the code.
so i would say detections is list of list with above format.
Hi, what is the format of the input to the tracker? is it: frame bbleft bbtop bbwidth bbheight confidence ?
I am not using any of the standard MOT detections. thanks.