[Fix] Fix SegTTAModel with no attribute '_gt_sem_seg' error #3152
Conversation
When using the - tta command for multi-scale prediction, and the test set is not annotated, although format_only has been set true in test_evaluator, but SegTTAModel class still threw error 'AttributeError: 'SegDataSample' object has no attribute '_gt_sem_seg''. The reason is SegTTAModel didn't determine if there were annotations in the dataset, so I added the code to make the judgment and let the program run normally on my computer.
|
I also encountered this bug.I changed the code with this method, but I can't see the progress bar now. envs/openmmlab/lib/python3.8/site-packages/mmengine/structures/pixel_data.py:83: UserWarning: The shape of value will convert from torch.Size([998, 998]) to torch.Size([1, 998, 998]) |
|
I would add that I can still see the progress bar.It is because there are too many warnings…… |
I haven't encountered this problem, perhaps you can try increasing the zoom ratio in tta_pipeline to avoid the image being too small. |
|
Hi @ZiAn-Su, |
|
Update and remove unnecessary code
Modify format
The new PR has been submitted. 😀 |
xiexinch
changed the title
…ab#3152) ## Motivation When using the - tta command for multi-scale prediction, and the test set is not annotated, although format_only has been set true in test_evaluator, but SegTTAModel class still threw error 'AttributeError: 'SegDataSample' object has no attribute '_gt_sem_seg''. ## Modification The reason is SegTTAModel didn't determine if there were annotations in the dataset, so I added the code to make the judgment and let the program run normally on my computer.
Motivation
When using the - tta command for multi-scale prediction, and the test set is not annotated, although format_only has been set true in test_evaluator, but SegTTAModel class still threw error 'AttributeError: 'SegDataSample' object has no attribute '_gt_sem_seg''.
Modification
The reason is SegTTAModel didn't determine if there were annotations in the dataset, so I added the code to make the judgment and let the program run normally on my computer.