isp-uv-es/mmflood-taco

MMFlood (SAR flood delineation)

This is a repackaging, not a new dataset. It is MMFlood (SAR flood delineation) by LINKS Foundation (Montello et al.), Copernicus Emergency Management Service, 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-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{montello2022mmflood,
  title   = {MMFlood: A Multimodal Dataset for Flood Delineation From Satellite Imagery},
  author  = {Montello, Fabio and Arnaudo, Edoardo and Rossi, Claudio},
  journal = {IEEE Access},
  volume  = {10},
  pages   = {96774--96787},
  year    = {2022},
  doi     = {10.1109/ACCESS.2022.3205419}
}

About the data

1748 Sentinel-1 tiles over 95 Copernicus EMS flood activations across Europe, each with a co-registered DEM and a binary flood-extent mask delineated by the EMS rapid mapping service. Converted from the original GeoTIFFs; tiles keep their native size and EPSG:4326 grid.

1,748 samples · splits: test 386 · train 1,242 · validation 120 · tasks: semantic-segmentation

Packaged as TACO v3.

Full description

Splits. The release's own, applied per activation so tiles of one event cannot straddle it: 1242 train, 120 val, 386 test.

Radiometry. Backscatter is linear power sigma-0, not dB. The pixel values are wholly positive, with VV averaging 0.196 and VH 0.047.

Labels. The mask's 255 marks pixels the authors exclude from scoring, where S1 swath data is missing, and is preserved as ignore_index.

Layers. MMFlood's OSM hydrography layer covers only 1,012 of the tiles and is not carried; zero-filling the rest would state absence of water where the release states nothing.

Getting started

git clone --recursive https://github.com/OscarPellicer/taco
pip install -e "taco/python[ml]"     # builds the reader: C++23, CMake, Ninja, pkg-config, libcurl >= 7.83, OpenSSL >= 3

Read it straight from the Hub:

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

path = hf_hub_download("isp-uv-es/mmflood-taco", "mmflood.zip", repo_type="dataset")
ds = Dataset(path)
plot_sample(ds[0])

or from a local copy:

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

Metadata without decoding anything:

import taco
taco.read("mmflood.zip")              # one Arrow table, levels joined

Samples

What a sample contains

role	slot	holds	modality	detail
input	sar	raster	sar	2 band(s), unit 1, physical
input	elevation	raster	elevation	1 band(s), unit m, physical
target	flood	mask	label_raster	2 classes

Licence

CC-BY-4.0

Providers: LINKS Foundation (Montello et al.), Copernicus Emergency Management Service

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.