search

Jun-CEN / Open-world-3D-semantic-segmentation

[ECCV 2022] Open-world Semantic Segmentation for LIDAR Point Clouds
69 stars 10 forks source link

Data download #8

Closed yifliu3 closed 2 years ago

yifliu3 commented 2 years ago

Hi, thanks for your great work!

I'd like to reproduce the results implemented in the paper, however, the dataset seems rather large, and can I just download part of the SemanticKitti and Nuscenes dataset?

yifliu3 commented 2 years ago

I find that in the build function, the code chooses the v1.0-test split of the whole part. So we only need to download this part, right?

if "nusc" in dataset_config['pc_dataset_type']:
        from nuscenes import NuScenes
        nusc = NuScenes(version='v1.0-test', dataroot=data_path, verbose=True)
Jun-CEN commented 2 years ago

Hi,

Thanks for your interest. If you want to reproduce the results, you should download the whole dataset because we use the whole dataset to train the model. If you want to evaluate the results, maybe the validation set or test set is enough.

yifliu3 commented 2 years ago

Hi,

Thanks for your prompt reply. So I also need to download the whole dataset for NuScenes, right?

Jun-CEN commented 2 years ago

Yes.

yifliu3 commented 2 years ago

Thanks a lot!