isp-uv-es/ossdd-taco

OSSDD - Open SAR Ship Detection Dataset

This is a repackaging, not a new dataset. It is OSSDD - Open SAR Ship Detection Dataset by OSSDD authors (Hammer et al.), ESA Copernicus / Sentinel-1, converted to TACO. Pixel values and labels are kept as released except where the description below says otherwise. All credit belongs to the original authors: if you use it, please cite them and follow their licence.

original dataset · paper · licence: CC-BY-NC-SA-4.0

Repackaged into TACO by the Image and Signal Processing Group (ISP), Universitat de València, within the ELLIOT project.

Citation

Please cite the original work:

@article{hammer2026ossdd,
  title   = {OSSDD -- a New Open Dataset for Sentinel-1 Ship Detection},
  author  = {Hammer, Horst and Hochstuhl, Sylvia and Thiele, Antje and Brosch, Tobias and Davidson, Padraig and Remiger, Tim and Teutsch, Michael},
  journal = {arXiv preprint arXiv:2608.01963},
  year    = {2026}
}

About the data

15161 Sentinel-1 GRD chips, dual polarisation, with every ship annotated as an axis-aligned box, as a rotated box given by its four corners, and as the release's own rasterised ship mask. A radar detection set confuses ships with speckle, wind streaks and the coastline rather than with shadows, and 1,978 of the chips are coastal for that reason.

15,161 samples · splits: test 3,036 · train 10,666 · validation 1,459 · tasks: object-detection, semantic-segmentation

Packaged as TACO v3.

Full description

Radiometry. Uncalibrated GRD amplitude, declared as digital numbers rather than as sigma-0. Negative values near strong reflectors are kept, the release shipping them.

Annotation. The two annotation files disagree about how many ships a chip holds on 1,251 chips, so both lists are carried with their own counts and the disagreement is flagged per sample rather than resolved by guess. 834 chips hold no ship at all and are kept as negatives.

Chip size. Follows the release's split: 700x700 for train, 512x512 for val and test (test 3036, train 10666, val 1459).

Left out. Captions ship with the release but are two templates keyed to the scene class, so no captioning task is declared. No geotransform is published and none is invented.

Getting started

git clone https://github.com/OscarPellicer/taco
pip install -e "taco/python[ml]"     # builds the native reader: needs CMake and Ninja

Read it straight from the Hub:

from huggingface_hub import snapshot_download
from taco.ml import Dataset, plot_sample

path = snapshot_download("isp-uv-es/ossdd-taco", repo_type="dataset")
ds = Dataset(path)                  # the repository is the container: parts + .tacocat
plot_sample(ds[0])

or from a local copy:

ds = Dataset("ossdd")
sample = ds[0]                      # {slot name: SlotValue}, arrays decoded
sample["vv"].array.shape

Metadata without decoding anything:

import taco
taco.read("ossdd/.tacocat")              # one Arrow table, levels joined

Samples

What a sample contains

role	slot	holds	modality	detail
input	vv	raster	sar	1 band(s), digital_number
input	vh	raster	sar	1 band(s), digital_number
target	boxes	bbox_2d
target	box_class	class_sequence		1 classes
target	boxes_rotated	polygon
target	ship_mask	mask	label_raster	2 classes

Licence

CC-BY-NC-SA-4.0

Terms of use:

• Non-commercial use only (CC-BY-NC-SA-4.0).
• Adapted material must be shared under CC-BY-NC-SA-4.0.

Required credits:

• Contains modified Copernicus Sentinel data, processed and annotated by the dataset authors. Annotations licensed under CC-BY-NC-SA-4.0.

Providers: OSSDD authors (Hammer et al.), ESA Copernicus / Sentinel-1

Acknowledgements

TACO was designed by César Aybar and is specified at https://asterisk.coop/taco/spec/.

Built by Oscar Pellicer within the ELLIOT project at the Image and Signal Processing Group (ISP), Universitat de València.