Documentation Index
Fetch the complete documentation index at: https://mintlify.com/dfki-ric/uxo-dataset2024/llms.txt
Use this file to discover all available pages before exploring further.
UXO Dataset 2024
Multimodal underwater unexploded ordnance perception dataset with acoustic sonar and optical camera data
About the Dataset
During the 20th century, millions of tons of munition were dumped into the oceans worldwide. After decades of decay, these unexploded ordnance (UXO) are causing significant problems. This dataset facilitates more efficient salvage efforts through autonomous underwater vehicles by providing representative multimodal data for acoustic and optical sensing of UXO underwater. The dataset includes close to 100 trajectories and over 74,000 frames of 3 distinct types of UXO recorded in a controlled environment using:- ARIS Explorer 3000 imaging sonar
- GoPro Hero 8 camera
- Custom-designed gantry crane for precise positioning
Access the Dataset
Download the dataset from Zenodo
View Data
Learn how to visualize synchronized recordings
Dataset Structure
Understand the data organization and formats
Processing Pipeline
Explore the preprocessing and export tools
Key Features
Multimodal Data
Synchronized ARIS sonar scans with matched GoPro footage
Accurate Transforms
Precise calibration matrices between sensors and targets
3D Models
Textured 3D models of UXO targets included
Annotations
Labeled camera frames with bounding boxes
Processing Tools
Complete preprocessing and export scripts
Open Access
Publicly available for research use
Dataset Properties
- Sonar scans of multiple UXO using ARIS Explorer 3000 imaging sonar
- Matched GoPro UHD frames for most sonar frames
- Labels for GoPro frames (multiply coordinates by 3)
- Known and accurate transforms between sonar and targets
- Known details of UXO targets including munition types, dimensions, and 3D models
- Tracked scan trajectories typical and achievable for non-experimental environments
- Publicly available at https://zenodo.org/records/13778485
Citation
When using this dataset or the processing code, please cite:Read the Paper
View the full research paper on ResearchGate