PASS: Point Assisted Sample Selection

This repository is based on [OpenPCDet].

Leveraging Anchor-based LiDAR 3D Object Detection via Point Assisted Sample Selection
Author: Shitao Chen, Haolin Zhang, Nanning Zheng
Paper is under review

Overview

• Changelog
• Model Zoo
• Installation
• Quick Demo
• Getting Started
• Acknowledgement

Changelog

[2022-03-04] PASS v0.1.0 is released.

Model Zoo

Waymo Open Dataset Baselines

anchor-based PV-RCNN++:tools/cfgs/waymo_models/pv_rcnn_plusplus_anchor_pass.yaml (20% training data)
anchor-based PASS-PV-RCNN++:tools/cfgs/waymo_models/pv_rcnn_plusplus_anchor_pass.yaml(20% training data)
anchor-based PASS-PV-RCNN++:tools/cfgs/waymo_models/pv_rcnn_plusplus_anchor_pass_whole.yaml(full training data)
PASS-PointPillars:tools/cfgs/waymo_models/pointpillar_1x_pass.yaml(20% training data)
PASS-PointPillars:tools/cfgs/waymo_models/pointpillar_1x_pass_whole.yaml(full training data)
PASS-SECOND:tools/cfgs/waymo_models/second_pass.yaml(20% training data)

We could not provide the above-pretrained models due to Waymo Dataset License Agreement, but you could easily achieve similar performance by training with the default configs. PASS is trained on one A800 GPU.

Acknowledgement

We would like to thank the authors of OpenPCDet for their open-source release of their codebase.

Citation

If you find this project useful in your research, please consider citing our work.