GACE

This is the demo code for the paper:

GACE: Geometry Aware Confidence Enhancement for Black-box 3D Object Detectors on LiDAR-Data
David Schinagl, Georg Krispel, Christian Fruhwirth-Reisinger Horst Possegger and Horst Bischof
IEEE/CVF International Conference on Computer Vision (ICCV), 2023
[Paper] [Supp.]

Overview

• Requirements
• Demo Example
• Acknowledgement
• Citation

Requirements

• Python 3.8+
• OpenPCDet (tested with version v0.6.0)
• PyTorch (tested with version 1.13.1)
• Pretrained SECOND Model on Waymo
• Waymo Dataset

Setup

1) OpenPCDet Installation

https://github.com/open-mmlab/OpenPCDet/blob/master/docs/INSTALL.md

2) Waymo Open Dataset Preparation

https://github.com/open-mmlab/OpenPCDet/blob/master/docs/GETTING_STARTED.md#waymo-open-dataset

3) SECOND Model Pre-Training

https://github.com/open-mmlab/OpenPCDet/blob/master/docs/GETTING_STARTED.md#train-a-model

GACE Demo

We provide a demo code showing the data extraction and training of a GACE model using a SECOND model as the base detector.

Run the demo as follows:

python gace-demo.py --ckpt ${PRETRAINED_SECOND_MODEL}

where ${PRETRAINED_SECOND_MODEL} is the path to the pretrained model weights.

Acknowledgement

We thank the authors of OpenPCDet for their open source release of their codebase.

Citation

If you find this code useful for your research, please cite

@InProceedings{Schinagl_2023_ICCV,
    author    = {Schinagl, David and Krispel, Georg and Fruhwirth-Reisinger, Christian and Possegger, Horst and Bischof, Horst},
    title     = {GACE: Geometry Aware Confidence Enhancement for Black-Box 3D Object Detectors on LiDAR-Data},
    booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV)},
    month     = {October},
    year      = {2023},
    pages     = {6566-6576}
}