Qibin Hou
Qibin Hou
Both of them do so.
Sorry for that, at present, only opencv3 is supported. You may modify the code yourself.
Yeah, sorry for that. At present, you may need to modify the code. I will fix this later on.
Some of the member functions of OpenCV2 are different from opencv3. You can try to compile the C++ part with gcc.
You can use torch.where() or get binary tensor directly using B_A = A > theta
You should ignore those regions in which attentioin values for some pair of classes are both large. For example, for location (i, j), A_{bike} = 0.7 and A_{person} = 0.8....
You can set regions with low saliency values (say < 32) to ignored regions. After all, the goal is to make the precision as high as possible. So, sometimes ignoring...
You email address? Or I can provide you with directly the specific numbers.
All branches are trained simultaneously. Sorry, I do not have a log file.