Skip to main content
The UXO Dataset 2024 is publicly available on Zenodo, providing multimodal synchronized data for acoustic and optical sensing of unexploded ordnance (UXO) underwater.

​
Dataset Overview

This dataset contains close to 100 trajectories and over 74,000 frames of 3 distinct types of UXO recorded in a controlled environment. It includes:
  • Sonar scans using an ARIS Explorer 3000 imaging sonar
  • Matched GoPro UHD frames for most sonar frames
  • Labeled camera frames with bounding box annotations
  • Accurate transforms between sonar and targets
  • 3D models of UXO targets with known dimensions and munition types
  • Realistic trajectories achievable in non-experimental environments
  • Calibration matrices and metadata
The dataset was recorded using an ARIS 3000 imaging sonar, a GoPro Hero 8, and a custom-designed gantry crane system.

​
Downloading from Zenodo

The complete dataset is hosted on Zenodo and can be accessed directly:

Download Dataset

Access the full UXO Dataset 2024 on Zenodo

​
Download Requirements

The dataset is substantial in size. Ensure you have sufficient storage space and a stable internet connection before downloading.
Before downloading, make sure you have:
  • Adequate disk space for the complete dataset
  • A reliable internet connection (recommended for large file downloads)
  • A tool for extracting compressed archives (if the dataset is provided in archive format)

​
What’s Included

The exported dataset uses simple, well-established formats for maximum compatibility:

Image Data

  • .pgm files for raw sonar frames
  • .png files for polar-transformed sonar images
  • .jpg files for camera frames

Metadata

  • .csv files for trajectories and frame metadata
  • .yaml files for configuration and calibration
  • .json files for structured data
  • .txt files for notes and documentation

​
Dataset Structure

After downloading and extracting, the dataset is organized by target type and recording: 
data_export/
└── recordings/
    └── <target_type>/
        └── <recording_folder>/
            ├── aris_raw/          # Raw sonar frames (.pgm)
            ├── aris_polar/        # Polar-transformed sonar (.png)
            ├── gopro/             # Camera frames (.jpg)
            ├── gantry.csv         # Trajectory data
            ├── aris_frame_meta.csv   # Frame-level metadata
            ├── aris_file_meta.yaml   # Recording-level metadata
            └── notes.txt          # Recording notes

​
File Formats Explained

  • aris_raw/: Raw sonar frames in PGM format as captured by the ARIS 3000
  • aris_polar/: Polar-transformed PNG images representing what the sonar was actually “seeing”
  • aris_frame_meta.csv: Frame-by-frame metadata including timestamps and sonar parameters
  • aris_file_meta.yaml: Recording-level configuration and settings
  • gopro/: Individual JPEG frames extracted from GoPro Hero 8 footage
  • GoPro frames are synchronized with sonar data
  • Label coordinates should be multiplied by 3 for correct scaling
  • gantry.csv: Contains the gantry crane trajectory and target poses
  • Indexed by ARIS frame index for easy synchronization
  • Includes accurate transforms between sonar and targets

​
Citation

When using this dataset, please cite the following paper: 
@INPROCEEDINGS{dahn2024uxo,
  author={Dahn, Nikolas and Firvida, Miguel Bande and Sharma, Proneet and 
          Christensen, Leif and Geisle, Oliver and Mohrmann, Jochen and 
          Frey, Torsten and Kumar Sanghamreddy, Prithvi and Kirchner, Frank},
  title={An Acoustic and Optical Dataset for the Perception of Underwater 
         Unexploded Ordnance (UXO)}, 
  booktitle={OCEANS 2024 - Halifax}, 
  year={2024},
  doi={10.1109/OCEANS55160.2024.10754316}
}

​
Additional Resources

​
Next Steps

After downloading the dataset:

Install Dependencies

Set up the required Python environment

View the Data

Learn how to visualize and explore the dataset