error when running av2 dataset
@LegendBC Hello, when running MapTR on the AV2 dataset. It goes into https://github.com/hustvl/MapTR/blob/main/projects/mmdet3d_plugin/datasets/av2_map_dataset.py#L627, then the training stops. Could you please provide some suggestions on solving the problem? Thanks.
Thanks for your remind, I will have a check to see if I can reproduce the error.
@outsidercsy How about https://github.com/hustvl/MapTR/blob/main/projects/mmdet3d_plugin/datasets/av2_map_dataset.py#L629? also comment it ?
And line 661 raises valueError.
Sorry. Add two failed logs here. https://github.com/hustvl/MapTR/blob/5a6bba1c1dd27c9c7851d7e2efb6939a91a4d86b/tools/data_converter/av2_converter.py#LL22C17-L22C17 '01bb304d-7bd8-35f8-bbef-7086b688e35e', '453e5558-6363-38e3-bf9b-42b5ba0a6f1d'
@outsidercsy So I have to append the FAIL_LOGShere and regenerate the pkl files for AV2?
More importantly, I have already removed the else conditions in both https://github.com/hustvl/MapTR/blob/main/projects/mmdet3d_plugin/datasets/av2_map_dataset.py#L660 and https://github.com/hustvl/MapTR/blob/main/projects/mmdet3d_plugin/datasets/av2_map_dataset.py#L626 and I am running the code without errors currently. Will there be any negative effects if I run in this way?
@outsidercsy So I have to append the
FAIL_LOGShere and regenerate the pkl files for AV2?More importantly, I have already removed the else conditions in both https://github.com/hustvl/MapTR/blob/main/projects/mmdet3d_plugin/datasets/av2_map_dataset.py#L660 and https://github.com/hustvl/MapTR/blob/main/projects/mmdet3d_plugin/datasets/av2_map_dataset.py#L626 and I am running the code without errors currently. Will there be any negative effects if I run in this way?
Sorry for the late reply. The code I wrote here is used to ensure the correctness and debug since the boundary in AV 2 should be polygon. There are some wrong cases as stated above. If you remove the wrong cases, the removed code will not influence the final results.
@outsidercsy So I have to append the
FAIL_LOGShere and regenerate the pkl files for AV2? More importantly, I have already removed the else conditions in both https://github.com/hustvl/MapTR/blob/main/projects/mmdet3d_plugin/datasets/av2_map_dataset.py#L660 and https://github.com/hustvl/MapTR/blob/main/projects/mmdet3d_plugin/datasets/av2_map_dataset.py#L626 and I am running the code without errors currently. Will there be any negative effects if I run in this way?Sorry for the late reply. The code I wrote here is used to ensure the correctness and debug since the boundary in AV 2 should be polygon. There are some wrong cases as stated above. If you remove the wrong cases, the removed code will not influence the final results.
@LegendBC Thank you.
To clarify again, now, I simply remove the two "else conditions" in the av2_map_dataset.py, and did not modify any other things of this repo's latest commit.
Now, the training is ongoing without errors, but I worry about the result of my training. Will it be lower than expected?
@outsidercsy So I have to append the
FAIL_LOGShere and regenerate the pkl files for AV2? More importantly, I have already removed the else conditions in both https://github.com/hustvl/MapTR/blob/main/projects/mmdet3d_plugin/datasets/av2_map_dataset.py#L660 and https://github.com/hustvl/MapTR/blob/main/projects/mmdet3d_plugin/datasets/av2_map_dataset.py#L626 and I am running the code without errors currently. Will there be any negative effects if I run in this way?Sorry for the late reply. The code I wrote here is used to ensure the correctness and debug since the boundary in AV 2 should be polygon. There are some wrong cases as stated above. If you remove the wrong cases, the removed code will not influence the final results.
@LegendBC Thank you.
To clarify again, now, I simply remove the two "else conditions" in the
av2_map_dataset.py, and did not modify any other things of this repo's latest commit.Now, the training is ongoing without errors, but I worry about the result of my training. Will it be lower than expected?
It wont deteriorate the final reaults. And in the next few days, we will upload our results and log for reference.