search

TRI-ML / dd3d

Official PyTorch implementation of DD3D: Is Pseudo-Lidar needed for Monocular 3D Object detection? (ICCV 2021), Dennis Park*, Rares Ambrus*, Vitor Guizilini, Jie Li, and Adrien Gaidon.
MIT License
451 stars 74 forks source link

how to train DD3D on custom datasets #10

Open williamhyin opened 2 years ago

williamhyin commented 2 years ago

Hello,

I only see the KITTI and Nuscense dataloader. Could you tell me how to build custom datasets which we can use to train DD3D. Or should we try to convert our custom dataset into KITTI or Nuscense format first, and then use their corresponding dataloader?

I am looking forward to your reply.

Thanks

dennis-park-TRI commented 2 years ago

Hello,

You can create your own dataset by writing a dataset that returns a list of dictionary with the keys and values match this interface: https://github.com/TRI-ML/dd3d/blob/main/tridet/data/dataset_mappers/dataset_mapper.py#L24

Please refer to the KITTI and nuScenes dataset classes for examples.