SpSequenceNet: Semantic Segmentation Network on 4D Point Clouds

PyTorch implementation of the paper "SpSequenceNet: Semantic Segmentation Network on 4D Point Clouds" in CVPR 2020. Vedio Link: here

Dependencies

Python >= 3.6

PyTorch >= 1.3

numpy >= 1.17.2

sparseconvnet >= 0.2

tqdm

Scripts

Training:

Firstly, the dataset setting is in the data_base and val_base of config.yaml. Modify it to the direction of your own dataset. Secondly, run as following:

cd train/semanticKITTI
python unet.py

Evaluation:

If you are validating your own trainined model, run as following:

cd train/semanticKITTI
python val_unet.py

If you want to use our trained model, add 'val_model_dir' under 'model' in the config.yaml. The val_model_dir is the directory of your model.

Our trained model is in here