Vegeta2020 / SE-SSD

SE-SSD: Self-Ensembling Single-Stage Object Detector From Point Cloud, CVPR 2021.
Apache License 2.0
811 stars 128 forks source link

dataset prep #75

Closed anran1231 closed 2 years ago

anran1231 commented 2 years ago

Hello, thanks for sharing the code. But I am very confused of dataset preparation. Could you please give more details about data organization? Best, Scarlett

Vegeta2020 commented 2 years ago

Hi, Scarlett. You may attend to the code in det3d/datasets/kitti and det3d/datasets/pipelines for grasping the overall framework of data preparation, there are several preprocesses like loading/augmentation/voxelization/assigntargets, they are simple but involve many code, so it's a matter of time and patience if you want to understand all. But if you want to do some minor modification, then you can revise the specific module directly as they are independent basically.