/Pitt-Radar

The repository is the release dataset of our Pitt-Radar of the CVPR 2023 paper "Azimuth Super-Resolution for FMCW Radar in Autonomous Driving"

Primary LanguagePythonMIT LicenseMIT

Pitt-Radar

This is a new radar dataset that contains 50k of radar scans collected using TI AWR1843BOOST in Pittsburgh, PA, USA. The RGB images are currently removed due to the internal privacy review. Only the Radar frames are provided currently.

Drag Racing

News

  • [Aug 2023] We release the 50k frames (radar maps only) at the link.

  • [March 2023] We release the first version of 5k frames of our dataset at the link in the resolution of 128 samples in range. Other parts of the dataset will be released later. Stay tuned.

Dataset Details

The dataset contains the following details:

  • 50k of raw radar scans in .bin format.

Dataset Format

The dataset is provided in , and includes the following:

  • A ./frames containing raw radar scans.

Data Collection

The radar scans and RGB camera images were collected using TI AWR1843BOOST in Pittsburgh, PA, USA. The data was collected by DENSO and includes scans from several different collection runs.

Usage

The dataset can be used for a variety of applications, including object detection and tracking, depth estimation, and scene understanding.

Citation

If you use this dataset in your research, please cite our paper:

@inproceedings{li2023pittradar,
  title={Azimuth Super-Resolution for FMCW Radar in Autonomous Driving},
  author={Li, Yu-Jhe and Hunt, Shawn and Park, Jinhyung and O'Toole, Matthew and Kitani, Kris M},
  booktitle={Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year={2023}
}

License

This dataset is released under the MIT license. Please see the LICENSE file for more details.

Contact

If you have any questions or feedback about the dataset, please contact us at yujheli@cs.cmu.edu.