kangming zhao

I have the same question while using data from indoor3d_sem_seg_hdf5_data

solved it is because of the version of pytorch, suggest using pytorch 1.0.0 so that you could ignore this error

this shows an error while you compile the tool about the pointnet++ part

this means we should learn how to compile the .c file, hope u could see the compile method in pointnet++ which also uses .c and .cuda files.

Excuse me， can u tell me how u trained using MobelNet40?

感觉你这里要么将图片改成300 * 300，要么去处理特征图的大小和计算

我感觉如果想去训练自己的模型需要把 model_dict.update(pretrained_dict) 这一句注释掉

我注释过之后我报错这个情况 Epoch:1/50 iter:0/3 || Loc_Loss: 3.2771 || Conf_Loss: 7.9131 || Traceback (most recent call last): File "马赛克/code/train.py", line 82, in loss_l, loss_c = criterion(out, targets) File "马赛克\torch\nn\modules\module.py", line 532, in...

我这里的conf函数确实是更改了，确实是我设置的的num_Classes,而且我也打印了看看

我找到原因了就是在你去计算conf_t的时候没有很好的计算好