search

ifzhang / FairMOT

[IJCV-2021] FairMOT: On the Fairness of Detection and Re-Identification in Multi-Object Tracking
MIT License
3.98k stars 936 forks source link

测试的时候dataloader读取摄像头，请问为何返回img和img0 #284

Open henbucuoshanghai opened 3 years ago

henbucuoshanghai commented 3 years ago 
    blob = torch.from_numpy(img).cuda().unsqueeze(0)
    online_targets = tracker.update(blob, img0)

img是img0缩小为6081088 原图img0是19201080 为何如此处理 为何吧img和img0塞进tracker 不懂？

ifzhang commented 3 years ago

img is the input of the model and we cannot directly put img0 to the model because the size is not suitable. When calcualating the bbox results, we need to calculate the bbox of img0, so we need the width and height of img0.