substantial difference between the test set and the validation set

[bqcs]

[bqcs]

I trained the model according to the original configuration in pointpillar.yaml, splitting the dataset into 80%-20% and 50%-50% for training and validation, respectively. Despite my efforts, when submitting the best results to the KITTI official for evaluation on the test set, I consistently fail to achieve metrics close to those reported in the original paper. The discrepancy is particularly noticeable for pedestrian and cyclist classes. Pointpillar：Pedestrian (3D Detection) | 52.08 % | 43.53 % | 41.49 % mine：Pedestrian (3D Detection) | 42.86 % | 36.09 % | 33.53 % What reasons could be causing this situation? thanks！ @sshaoshuai

[github-actions[bot]]

This issue is stale because it has been open for 30 days with no activity.

[github-actions[bot]]

This issue was closed because it has been inactive for 14 days since being marked as stale.