Qiang Hu
Results
3
issues of
Qiang Hu
error
1
one of the variables needed for gradient computation has been modified by an inplace operation: [torch.cuda.FloatTensor [2, 128, 80, 88]], which is output 0 of ReluBackward0, is at version 3;...
question
1
请问boxlevelset在mmdet框架中测试，如何只输出掩膜图？就是不需要有原图的叠加
请问您在文章中，这两个指标的P,R,F1分别是怎么计算的？PraNet等是分割模型，是如何计算Detection的P,R,F1的？