paul007pl
commented
2 years ago EMD loss requires the "input point cloud" and the "GT point cloud" to have the same number of points. You may check the emd loss here: http://robotics.stanford.edu/~rubner/papers/rubnerIjcv00.pdf
Traceback (most recent call last): File "/home/zhao/doc/MVP_Benchmark-main/completion/train.py", line 219, in
train()
File "/home/zhao/doc/MVP_Benchmark-main/completion/train.py", line 140, in train
out2, loss2, net_loss = net(inputs, gt, alpha=alpha)
File "/home/zhao/anaconda3/envs/mvp/lib/python3.7/site-packages/torch/nn/modules/module.py", line 550, in call
result = self.forward(*input, kwargs)
File "/home/zhao/anaconda3/envs/mvp/lib/python3.7/site-packages/torch/nn/parallel/data_parallel.py", line 153, in forward
return self.module(*inputs[0], *kwargs[0])
File "/home/zhao/anaconda3/envs/mvp/lib/python3.7/site-packages/torch/nn/modules/module.py", line 550, in call
result = self.forward(input, kwargs)
File "/home/zhao/doc/MVP_Benchmark-main/completion/models/pcn.py", line 96, in forward
loss1 = calc_emd(out1, gt)
File "/home/zhao/doc/MVP_Benchmark-main/completion/model_utils.py", line 83, in calcemd
dist, = emd_loss(output, gt, eps, iterations)
File "/home/zhao/anaconda3/envs/mvp/lib/python3.7/site-packages/torch/nn/modules/module.py", line 550, in call
result = self.forward(*input, **kwargs)
File "../utils/metrics/EMD/emd_module.py", line 88, in forward
return emdFunction.apply(input1, input2, eps, iters)
File "../utils/metrics/EMD/emd_module.py", line 47, in forward
assert(n == m)
AssertionError
进程已结束,退出代码1