liaozhihao2757/transrvnet

Paper is Under Review

TransRVNet

Code coming soon...Currently we only show our results for visualization and evaluation purposes.

First please download the SemanticPOSS Dataset from the official website.

Then download our predictions from Google Drive.

Visualize Example:

• Visualize GT:

  python visualize.py -d /dataset -s sequences:00-06

• Visualize Our Predictions:

  python visualize.py -d /dataset -p /predictions -s 02/03

• In the visualization:

  To navigate:
  
      b: back (previous scan)
  
      n: next (next scan)
  
      q: quit (exit program)
  

Visualize Video:

Eval Example:

• Eval Seq 02:

  python evaluate_iou.py -d /dataset -p /predictions -s valid02

• Eval seq 03:

  python evaluate_iou.py -d /dataset -p /predictions -s valid03

The following results will be obtained:

Acknowledgment

The code is partly based on LiDAR-Bonnetal and SalsaNext. Thanks for their open source work.

Citation

Currently, please consider citing:

@inproceedings{pan2020semanticposs,
  title={Semanticposs: A point cloud dataset with large quantity of dynamic instances},
  author={Pan, Yancheng and Gao, Biao and Mei, Jilin and Geng, Sibo and Li, Chengkun and Zhao, Huijing},
  booktitle={2020 IEEE Intelligent Vehicles Symposium (IV)},
  pages={687--693},
  year={2020},
  organization={IEEE}
}