Nengzhong Yin
Nengzhong Yin
if i use motion and reid, i should replace multiracker.py into tracker.py and set --match_thres 0.4 but how about conf_thres?
i find the spconv in Querydet is based on spconv 1.x ,but i setup spconv 2.x in my environment, i have some bug when i run the code, i don`t...
this from querydet,it uses spconv,when i infer,there are some questions about spcov。 
Thanks for your code! @ifzhang when i use the pretrained model hrnetv2_w18_imagenet_pretrained.pth or w32 and change the arch into hrnet_18 for training, it will show the error KeyError: 'epoch' 'state_dict',...
win10下运行compile.sh会报这个错/(ㄒoㄒ)/
Thank for your great work!i want to evaluate in visdrone? Can i request trained model? my email is : [email protected] thanks!
when i train in visdrone, i just modify the batchsize and the max iter ,but there is a promblem in 768iter 
i only see ReID is used in bytetrack? Does ocsort also can use ReID?
When using MultiDatasetsEvaluator, I find that it will show that the evaluator shows no cumulative_sizes making it impossible to proceed, but my val_dataloader uses concatdatasets. Here is my config: val_evaluator...