wrong dimensions of gts in 3D object detection evaluation

argoverse / av2-api

Argoverse 2: Next generation datasets for self-driving perception and forecasting.

https://argoverse.github.io/user-guide/

MIT License

317 stars 75 forks source link

wrong dimensions of gts in 3D object detection evaluation #246

Closed chris-23 closed 7 months ago

chris-23 commented 7 months ago

When I follow the 3D object detection tutorial it seems the ground truth annotations don't contain a _loginfo colmn. If I run it I get this error message: KeyError: 'log_id' According to the documentation in eval.py the evaluate function expects a pandas dataframe of dimension Mx15, wheras only Mx14 gets passed, after loading the annotations with gts = read_all_annotations(dataset_dir=dataset_dir, split='val') . So I assume a log_info colums needs to be added in this function?

benjaminrwilson commented 7 months ago

Hi @chris-23,

Thanks for bringing this up. I've added a log_id in that function in this PR: https://github.com/argoverse/av2-api/pull/247.

Let us know if you have any other questions!