isp-uv-es/spacenet7-taco

SpaceNet 7 Multi-Temporal Urban Development

This is a repackaging, not a new dataset. It is SpaceNet 7 Multi-Temporal Urban Development by Planet Labs, SpaceNet LLC / CosmiQ Works, 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-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:

@inproceedings{vanetten2021multitemporal,
  title     = {The Multi-Temporal Urban Development SpaceNet Dataset},
  author    = {Van Etten, Adam and Hogan, Daniel and Manso, Jesus Martinez and Shermeyer, Jacob and Weir, Nicholas and Lewis, Ryan},
  booktitle = {CVPR},
  year = {2021}
}

About the data

Building footprints through time: 60 areas of interest, each a monthly Planet visual mosaic series of 18-26 frames at about 4.8 m over 2018-2020, 1423 frames in all.

60 samples · splits: train 48 · validation 12 · tasks: semantic-segmentation, change-detection

Packaged as TACO v3.

Full description

Target. A binary building mask for every frame. This distribution carries the rasterised masks only, not the per-building polygons with tracking ids.

Splits. The challenge's held-out test AOIs, whose labels are withheld, are in the held-out half below. The split here is not the publisher's: it is by AOI.

Held-out half

spacenet7_test.zip holds the 20 samples whose targets the publisher withheld. Same inputs, no target slots: it is there to be predicted on and submitted, and it belongs in no training mixture.

import os
from huggingface_hub import snapshot_download
from taco.ml import Dataset

root = snapshot_download("isp-uv-es/spacenet7-taco", repo_type="dataset",
                         allow_patterns=["spacenet7_test.zip", "spacenet7_test.zip/*", "spacenet7_test.zip/.tacocat/*"])
ds = Dataset(os.path.join(root, "spacenet7_test.zip"))

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/spacenet7-taco", "spacenet7.zip", repo_type="dataset")
ds = Dataset(path)
plot_sample(ds[0])

or from a local copy:

ds = Dataset("spacenet7.zip")
sample = ds[0]                      # {slot name: SlotValue}, arrays decoded
sample["image"].array.shape

Metadata without decoding anything:

import taco
taco.read("spacenet7.zip")              # one Arrow table, levels joined

Samples

What a sample contains

role	slot	holds	modality	detail
input	image	raster_series	optical	3 band(s), render
target	buildings	mask_set	label_raster

Licence

CC-BY-SA-4.0

Terms of use:

• Adapted material must be shared under CC-BY-SA-4.0.

Required credits:

• The SpaceNet Dataset by SpaceNet Partners is licensed under CC-BY-SA-4.0
• Imagery: Planet Labs

Providers: Planet Labs, SpaceNet LLC / CosmiQ Works

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.