Datasets:

Languages:
English
ArXiv:
Libraries:
Datasets
License:
Dataset Viewer (First 5GB)
Auto-converted to Parquet Duplicate
Search is not available for this dataset
image
imagewidth (px)
1.92k
3.84k
End of preview. Expand in Data Studio

DeepSea MOT

DeepSea MOT is a benchmark dataset for multi-object tracking on deep-sea video.

Dataset Description

DeepSea MOT consists of 4 video sequences (2 midwater, 2 benthic) with a total of 2,400 frames and 57,376 annotated objects comprising 188 tracks. The videos were captured by the Monterey Bay Aquarium Research Institute (MBARI) using remotely operated vehicles (ROVs) Doc Ricketts and Ventana in deep-sea environments, showcasing a variety of marine species and underwater scenes.

[2025-12-08] New 4K Sequences: In addition to the original sequences, DeepSea MOT now includes two new 4K resolution sequences (Benthic Simple and Midwater Simple) to facilitate research on high-resolution underwater video analysis.

[2026-02-08] New Marine Debris Sequences: DeepSea MOT now includes five marine debris sequences to facilitate detection of debris in underwater video. Three video sequences were collected by MBARI (MD_ROP, MD_CAN, and MD_CLS) and two sequences were collected by IFREMER (MD_BOT and MD_FLN).

Note: these sequences use a different class ID mapping than the original DeepSea MOT sequences.

File Structure

The dataset is organized as follows:

data/
β”œβ”€β”€ vidseq_names.txt          # TrackEval sequence names file
β”œβ”€β”€ BD/                       # Benthic Difficult sequence
β”‚   β”œβ”€β”€ BD.mov                # Source video file
β”‚   β”œβ”€β”€ gt.txt                # MOT Challenge format ground truth
β”‚   β”œβ”€β”€ seqinfo.ini           # TrackEval sequence information
β”‚   β”œβ”€β”€ images/               # Frame images (JPG format)
β”‚   β”‚   β”œβ”€β”€ BD_001.jpg
β”‚   β”‚   β”œβ”€β”€ BD_002.jpg
β”‚   β”‚   └── ...
β”‚   β”œβ”€β”€ labels/               # YOLO-formatted annotation files (TXT)
β”‚   └── xml/                  # Pascal VOC annotation files (from RectLabel)
β”œβ”€β”€ BS/                       # Benthic Simple sequence
β”‚   β”œβ”€β”€ BS.mov
β”‚   β”œβ”€β”€ gt.txt
β”‚   β”œβ”€β”€ seqinfo.ini
β”‚   β”œβ”€β”€ images/
β”‚   β”œβ”€β”€ labels/
β”‚   └── xml/
β”œβ”€β”€ MWD/                      # Midwater Difficult sequence
β”‚   β”œβ”€β”€ MWD.mov
β”‚   β”œβ”€β”€ gt.txt
β”‚   β”œβ”€β”€ seqinfo.ini
β”‚   β”œβ”€β”€ images/
β”‚   β”œβ”€β”€ labels/
β”‚   └── xml/
β”œβ”€β”€ MWS/                      # Midwater Simple sequence
|   β”œβ”€β”€ MWS.mov
|   β”œβ”€β”€ gt.txt
|   β”œβ”€β”€ seqinfo.ini
|   β”œβ”€β”€ images/
|   β”œβ”€β”€ labels/
|   └── xml/
β”œβ”€β”€ ...

Each sequence directory contains:

  • Source video (.mov): Original ROV footage
  • Ground truth (gt.txt): MOT Challenge format annotations for tracking evaluation
  • Sequence info (seqinfo.ini): Metadata file for TrackEval compatibility
  • Images (images/): Individual frame extractions in JPG format
  • YOLO labels (labels/): Object detection annotations in YOLO format (TXT files)
  • Pascal VOC (xml/): Object detection annotations in Pascal VOC format (generated via RectLabel)

Additional Information

Dataset Curators

Authors of [1]:

  • Kevin Barnard
  • Elaine Liu
  • Kristine Walz
  • Brian Schlining
  • Nancy Jacobsen Stout
  • Lonny Lundsten

Citation Information

@article{barnard2025deepseamot,
    author = {Barnard, Kevin and Liu, Elaine and Walz, Kristine and Schlining, Brian and Stout, Nancy Jacobsen and Lundsten, Lonny},
    title = { {DeepSea MOT}: A benchmark dataset for multi-object tracking on deep-sea video},
    year = {2025},
    journal = {arXiv preprint arXiv:2509.03499},
    doi = {10.48550/arXiv.2509.03499},
}
Downloads last month
578

Paper for MBARI-org/DeepSea-MOT