Mini-version of Nuscenes Dataset

[SMSajadi99]

Hello everyone!

Tnx for sharing your helpful project.

I have started working on your Github. However, I want to run the code with mini-version of the Nuscenes dataset. In all part of your code, only the full dataset is used. Since, I have three questions!

1- I want to know that, is it possible to run the code with mini-version?
2- Also, is it possible to generate the .pkl files with mmdetection? Or, can you provide it on your Github?
3- Are sweeps data (I mean non-annotated frames) necessary for your training algorithm? Or, just by the keyframes we can run the training code.

Thank you.

[kaixinbear]

Thanks for your questions.
Since I haven't done any experiment on mini-version, please feel free to ask if you come cross bugs.
1- CAPE could be run on mini-version. I comprehend mini-version of nuscenes as a subset of the whole dataset. So just prepare the data format as the full dataset.
2- I think you just need to place v1.0-mini under the nuscenes root and run the script:
python tools/create_data.py nuscenes --root-path ./data/nuscenes --out-dir ./data/nuscenes --extra-tag nuscenes --version v1.0-mini
3- The sweeps data are not necessary but you need to generate the custom pkl file and modify the dataset by yourself.
Or for simplicity, the provided pkl file has included the keyframes. You can specify the keyframe id for just the keyframes.