@CaptainEven
请问,对于有的类别制作检测的问题,我没有搞懂这个问题
如果已经训练了一个30个类别的模型,只对其中的0,4,6做跟踪其他的做检测,那么只需要在reid_cls_ids中设置0,4,6就可以吗,我这样设置的,然后现实这个错误:
Fix size testing.
training chunk_sizes: [2, 3, 3, 2]
The output will be saved to /home/sunyue/MCMOT-master/src/lib/../../exp/mot/default13
Net input image size: 1088×608
[Err]: configuration conflict of reid_cls_ids and num_classes!
Traceback (most recent call last):
File "demo.py", line 219, in
run_demo(opt)
File "demo.py", line 41, in run_demo
result_root = opt.output_root if opt.output_root != '' else '.'
AttributeError: 'NoneType' object has no attribute 'output_root'
@CaptainEven 请问,对于有的类别制作检测的问题,我没有搞懂这个问题 如果已经训练了一个30个类别的模型,只对其中的0,4,6做跟踪其他的做检测,那么只需要在reid_cls_ids中设置0,4,6就可以吗,我这样设置的,然后现实这个错误: Fix size testing. training chunk_sizes: [2, 3, 3, 2] The output will be saved to /home/sunyue/MCMOT-master/src/lib/../../exp/mot/default13 Net input image size: 1088×608 [Err]: configuration conflict of reid_cls_ids and num_classes! Traceback (most recent call last): File "demo.py", line 219, in
run_demo(opt)
File "demo.py", line 41, in run_demo
result_root = opt.output_root if opt.output_root != '' else '.'
AttributeError: 'NoneType' object has no attribute 'output_root'
还是在训练的时候这么设置,在训练的时候设置也会出错。