作者你好 我在torch1.11上使用roi-align出现RuntimeError

Traceback (most recent call last):
  File "c:/Users/29006/Desktop/GAR/scripts/train_volleyball_stage1.py", line 21, in <module>
    train_net(cfg)
  File ".\train_net.py", line 110, in train_net
    return forward_call(*input, **kwargs)
  File ".\base_model.py", line 117, in forward
    boxes_idx_flat)  #B*T*N, D, K, K,
  File "D:\Anaconda\envs\GCN\lib\site-packages\torch\nn\modules\module.py", line 1110, in _call_impl
    return forward_call(*input, **kwargs)
  File "D:\Anaconda\envs\GCN\lib\site-packages\roi_align-0.0.2-py3.7-win-amd64.egg\roi_align\roi_align.py", line 48, in forward
    return CropAndResizeFunction.apply(featuremap, boxes, box_ind, self.crop_height, self.crop_width, self.extrapolation_value)
  File "D:\Anaconda\envs\GCN\lib\site-packages\roi_align-0.0.2-py3.7-win-amd64.egg\roi_align\crop_and_resize.py", line 25, in forward        
    ctx.extrapolation_value, ctx.crop_height, ctx.crop_width, crops)
RuntimeError: image.is_contiguous()INTERNAL ASSERT FAILED at "C:\\Users\\29006\\Desktop\\RoIAlign\\roi_align\\src\\crop_and_resize_gpu.cpp":27, please report a bug to PyTorch. image must be contiguous

请问你在使用RoIAlign-pytorch时是否也遇到了这样的错误呢？请问我应该从什么方向解决它？

Hello,

Thank you very much for the paper and library. I'm trying to reproduce the results using the suggested training script in the readme file.

I'm getting the following errors: File "/home/ec2-user/GroupFormer/main.py", line 56, in <module> main() File "/home/ec2-user/GroupFormer/main.py", line 53, in main group_helper.train() File "/home/ec2-user/GroupFormer/group/group.py", line 239, in train self.train_epoch() File "/home/ec2-user/GroupFormer/group/group.py", line 259, in train_epoch actions_loss, activities_loss, aux_loss, loss = self.forward(batch) File "/home/ec2-user/GroupFormer/group/group.py", line 200, in forward actions, activities, aux_loss = self.model(batch[0], batch[1], batch[4], batch[5]) File "/home/ec2-user/.local/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/ec2-user/GroupFormer/group/utils/distributed_utils.py", line 18, in forward return self.module(*inputs, **kwargs) File "/home/ec2-user/.local/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/ec2-user/GroupFormer/group/models/__init__.py", line 149, in forward actions_scores1, activities_scores1, aux_loss1 = self.head(boxes_features, global_token) File "/home/ec2-user/.local/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/ec2-user/GroupFormer/group/models/head/st_plus_tr_cross_cluster.py", line 208, in forward group = self.group_tr(group_query,x).reshape(1,B*T,-1) File "/home/ec2-user/.local/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/ec2-user/GroupFormer/group/models/transformer.py", line 104, in forward output = layer(output, memory, tgt_mask=tgt_mask, File "/home/ec2-user/.local/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/ec2-user/GroupFormer/group/models/transformer.py", line 243, in forward return self.forward_pre(tgt, memory, memory_mask, memory_key_padding_mask, pos) File "/home/ec2-user/GroupFormer/group/models/transformer.py", line 227, in forward_pre tgt2 = self.self_attn(q, k, value=src, attn_mask=src_mask, File "/home/ec2-user/.local/lib/python3.9/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/ec2-user/.local/lib/python3.9/site-packages/torch/nn/modules/activation.py", line 1153, in forward attn_output, attn_output_weights = F.multi_head_attention_forward( File "/home/ec2-user/.local/lib/python3.9/site-packages/torch/nn/functional.py", line 5030, in multi_head_attention_forward is_batched = _mha_shape_check(query, key, value, key_padding_mask, attn_mask, num_heads) File "/home/ec2-user/.local/lib/python3.9/site-packages/torch/nn/functional.py", line 4874, in _mha_shape_check assert key.dim() == 3 and value.dim() == 3, \ AssertionError: For batched (3-D) "query", expected "key" and "value" to be 3-D but found 4-D and 4-D tensors respectively

Any ideas why does it happen?

Thanks in advance!

Hello, I found in the test log that the type of head and pose_head is ST_PLUS_TR_cross_cluster_v2, but there is no ST_PLUS_TR_cross_cluster_V2 module in the code. Have you made any changes based on ST_PLUS_TR_cross_cluster? I used the checkpoint provided by you and only got the results of Action Accuracy 83.988% and Activity Accuracy 94.914%, which are quite different from the results provided by you

File "/content/GroupFormer/group/utils/distributed_utils.py", line 73, in dist_init_pytorch rank = int(os.environ['RANK']) File "/usr/lib/python3.7/os.py", line 681, in getitem raise KeyError(key) from None KeyError: 'RANK

Thank you for the great work! I tried to reproduce the result of Collective Activity Dataset but was able to achieve only about 88% accuracy. If possible, can you provide the configuration file and pre-trained model for Collective Activity Dataset?

Hi, thanks for your work and sharing your codes!

When I tried to test your uploaded model performance and unzip the checkpoint.pth.tar, I occurred to an error as follows:

$ tar -xvf checkpoint.pth.tar
tar: This does not look like a tar archive
tar: Skipping to next header
tar: Exiting with failure status due to previous errors

I wander whether you get the same error, and I hope you give me some help. I know this tar file is as huge as about 500 MB, could I trouble you to upload the unzipped checkpoint.pth file to Google Driver directly again? Thank you in advance.

Regard