Object-specific Distance Dataset

Datasets constructed based on KITTI and NuScenes object detection datasets. For each object in the RGB image, we calculate a object-specific distance (in meters) from a set of LiDAR point clouds. For more details, please refer to our paper.

drawing

Citation

If you are using any part of our Object-specific Distance dataset(s) or codes, please add a reference to our paper. All the original data (i.e., object annoations, RGB images, LiDAR point clouds) is from KITTT or NuScenes datasets. Please cite their paper(s) as well.

@inproceedings{zhu2019learning,
title={Learning object-specific distance from a monocular image},
author={Zhu, Jing and Fang, Yi},
booktitle={Proceedings of the IEEE International Conference on Computer Vision},
pages={3839–3848},
year={2019}
}

License

Please notice that this dataset is made available for academic research purpose only. The copyright of the original data (i.e., object annotation, images, LiDAR point clouds, calibration settings) belongs to the original owners (KITTI or NuScenes team).

Download

  • KITTI-based
    • Overview:

      image: RGB images (collect from KITTI left color images)
      distLabel: a label for each image. Please readme.txt for full description.
      project3DTo2D: 4x3 projection matrix used to project 3D (XYZ) keypoint to 2D (u,v) image plane

    • Distance Dataset
    • Code to calculate distance annotation
  • NuScenes-based
    • Distance Dataset (to be released)