Skip to content
chrisyanPublic

About

[IEEE IV 2025] Code and dataset - "Framework and Multi-modal Dataset for Roadwork Zone Detection and Geo-localization"

Resources

Stars

2 stars

Watchers

1 watching

Forks

Latest commit

 

History

7 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

RZDG

Code and dataset of paper IEEE IV 2025 "Framework and Multi-modal Dataset for Roadwork Zone Detection and Geo-localization"

The RZDG (Roadwork Zone Detection and Geo-localization) dataset is a multi-modal dataset designed to support research in roadwork zone perception, detection, tracking and geo-localization. It contains both real-world (RZDG-real) and simulated (RZDG-sim) data, structured in the KITTI format, and supports tasks such as:

  • Image semantic segmentation
  • 3D object detection and tracking
  • Geo-localization of roadwork zones

The download is available at: RZDG Download Link

Dataset Structure

RZDG-Sim (Simulated)

  • Objects: 419 road beacons, 254 barriers
  • Scenarios: 203 distinct roadwork zone scenarios across 8 simulated towns
  • Frames: 8591

RZDG-Real (Real-World)

  • Objects: 3643 road beacons, 8837 barriers
  • Scenarios: 28 distinct roadwork zone scenarios
  • Frames: 1357

The overall structure of the dataset is as follows:

RWDG_dataset
├── ImageSets
├── kitti_gt_database
├── testing
│   ├── calib
│   ├── image_0
│   └── lidar
└── training
    ├── calib
    ├── image_0
    ├── label_0
    └── lidar

Detection & Tracking Pipeline

We adopt MMDetection3D for 3D object detection and AB3DMOT for 3D object tracking.

To adapt these tools to the RZDG dataset, we provide custom modifications in the form of patch files located in the patches/ directory.

File Target Repository Description
patches/mmdetection3d.patch MMDetection3D Modifications to support RZDG dataset format, object classes, and evaluation setup
patches/AB3DMOT.patch AB3DMOT Tracker changes to handle geo-localization and RZDG-specific object cues

Use following commands to apply patches to the Original Repositories

cd path/to/mmdetection3d
git apply ../patches/mmdetection3d.patch

cd path/to/AB3DMOT
git apply ../patches/AB3DMOT.patch

Acknowledgment

This project is built upon the excellent open-source frameworks MMDetection3D and AB3DMOT. We sincerely thank the developers and maintainers of these repositories for their contributions to the research community.

Cite our paper

@inproceedings{yan2025framework,
  title={Framework and Multi-modal Dataset for Roadwork Zone Detection and Geo-localization},
  author={Yan, Zhiran and Xin, Yutong and Shenoi, S Shyam and Song, Rui and Elger, Gordon},
  booktitle={2025 IEEE Intelligent Vehicles Symposium (IV)},
  pages={1758--1765},
  year={2025},
  organization={IEEE}
}

About

[IEEE IV 2025] Code and dataset - "Framework and Multi-modal Dataset for Roadwork Zone Detection and Geo-localization"

Resources

Stars

2 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors