RuntimeError: expected input[100, 3, 512, 32] to have 131 channels

[pramishp]

I am trying to run the training on my Ubuntu 18.04.5 with python=3.7.10, torch=1.1.0. However, I am getting the following error when trying to run training with batchsize=100

......
  File "VoteHMR/src/models/backbone_module.py", line 95, in forward
    xyz, features, fps_inds = self.sa2(xyz, features)  # this fps_inds is just 0,1,...,1023
  File "/home/pramish/miniconda3/envs/votehmr/lib/python3.7/site-packages/torch/nn/modules/module.py", line 493, in __call__
    result = self.forward(*input, **kwargs)
  File "VoteHMR/src/pointnet2/pointnet2_modules.py", line 253, in forward
    grouped_features
  File "/home/pramish/miniconda3/envs/votehmr/lib/python3.7/site-packages/torch/nn/modules/module.py", line 493, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/pramish/miniconda3/envs/votehmr/lib/python3.7/site-packages/torch/nn/modules/container.py", line 92, in forward
    input = module(input)
  File "/home/pramish/miniconda3/envs/votehmr/lib/python3.7/site-packages/torch/nn/modules/module.py", line 493, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/pramish/miniconda3/envs/votehmr/lib/python3.7/site-packages/torch/nn/modules/container.py", line 92, in forward
    input = module(input)
  File "/home/pramish/miniconda3/envs/votehmr/lib/python3.7/site-packages/torch/nn/modules/module.py", line 493, in __call__
    result = self.forward(*input, **kwargs)
  File "/home/pramish/miniconda3/envs/votehmr/lib/python3.7/site-packages/torch/nn/modules/conv.py", line 338, in forward
    self.padding, self.dilation, self.groups)
RuntimeError: Given groups=1, weight of size 128 131 1 1, expected input[100, 3, 512, 32] to have 131 channels, but got 3 channels instead

[hanabi7]

thanks you for your issue, this is caused by my personal mistake when uploading the backbone module files, you can check the newest commit. that should fix the errors

[pramishp]

Thanks for the fix. However, in this line of the backbone net you are passing gt_segmentation. I don't think that is intended. Also, the shape of the gt_segment doesn't match the input param shape of the labels param in the ProposalModule forward call.

If possible can you please provide me with your email?

[hanabi7]

my email is @.***

	刘冠泽

@.*** |

---- 回复的原邮件 ---- | 发件人 | @.> | | 日期 | 2022年06月23日 18:06 | | 收件人 | @.> | | 抄送至 | @.**@.> | | 主题 | Re: [hanabi7/VoteHMR] RuntimeError: expected input[100, 3, 512, 32] to have 131 channels (Issue #3) |

Thanks for the fix. However, in this line of the backbone net you are passing gt_segmentation. I don't think that is intended. Also, the shape of the gt_segment doesn't match the input param shape of the labels param in the ProposalModule forward call.

If possible can you please provide me with your email?

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

[hanabi7]

please letme know your email address for easier communication.

	刘冠泽

@.*** |

---- 回复的原邮件 ---- | 发件人 | @.> | | 日期 | 2022年06月23日 18:06 | | 收件人 | @.> | | 抄送至 | @.**@.> | | 主题 | Re: [hanabi7/VoteHMR] RuntimeError: expected input[100, 3, 512, 32] to have 131 channels (Issue #3) |

Thanks for the fix. However, in this line of the backbone net you are passing gt_segmentation. I don't think that is intended. Also, the shape of the gt_segment doesn't match the input param shape of the labels param in the ProposalModule forward call.

If possible can you please provide me with your email?

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: @.***>

[daixiangzi]

please letme know your email address for easier communication. | | 刘冠泽 | | @.*** | … your problem is solved?

[daixiangzi]

you reproduce this paper?

[daixiangzi]

Thanks for the fix. However, in this line of the backbone net you are passing gt_segmentation. I don't think that is intended. Also, the shape of the gt_segment doesn't match the input param shape of the labels param in the ProposalModule forward call.

If possible can you please provide me with your email?

same problem

[daixiangzi]

代码中有很多对不齐的地方。。。。。。