Find research datasets worth reusing

Search datasets from major research repositories and use ShareScore to quickly assess how well each record supports discovery, access, and reuse.

528

datasets available to search

ShareScore release 0.7.1

Dataset results

528 results for “Land cover”

Learn how ShareScore rates datasets ↗

dryad32/100

Preserving the woody plant tree of life in China under future climate and land-cover changes

The tree of life (TOL) is severely threatened by climate and land-cover changes. Preserving the TOL is urgent, but has not been included in the post-2020 global biodiversity framework. Protected areas (PAs) are fundamental for biological conservation. However, we know little about the effectiveness of existing PAs in preserving the TOL of plants and how to prioritize PA expansion for better TOL preservation under future climate and land-cover changes. Here, using high-resolution distribution maps of 8732 woody species in China and phylogeny-based Zonation, we find that current PAs perform poorly in preserving the TOL both at the present and in the 2070s. The geographical coverage of TOL branches by current PAs is ca. 9%, and < 3% of the identified priority areas for preserving the TOL are currently protected. Interestingly, the geographical coverage of TOL branches by PAs will be improved from 9% to 52–79% by the identified priority areas for PA expansion. Human pressures in the identified priority areas are high, leading to high costs for future PA expansion. We thus suggest that besides nature reserves and national parks, other effective area-based conservation measures should be considered. Our study argues for the inclusion of preserving the TOL in the post-2020 conservation framework and provides references for decision-makers to preserve the Earth's evolutionary history.

opencc-zeroNov 2022View details →

zenodo32/100

Represent project UC2 Land Cover Dataset

The dataset is composed of Sentinel-2 data, while the ground truths come from Copernicus Land Monitoring Service – Hot Spot Mapping (HSM), that provides high resolution land cover maps over several Natural Protected Areas mainly in Africa. The land cover legend is based on the FAO Land Cover Classification System (LCCS). These labels have been produced and validated mainly through photointerpretation of HR data. The mapped area of interest (AOI) represents a key landscape for conservation area (KLC). The KLC has a total size of almost 14,000,000 140,000 km2 and is covering a vast area in southern Tanzania all the way to northern Mozambique. In Tanzania the AOI is covering the entire Selous game reserve, an area of around 50,000 km2, representing 6% of Tanzania’s land surface. This world heritage site is not only the oldest, but also the largest single protected area in Africa. It is characterised by an extensive area of natural miombo woodlands representing also one of the most extensive forest areas under protection. the reserve contains some of the most important populations of elephants, buffalos, antelopes, lions, wild dogs and other predators in Africa. Located in northern Mozambique, the Niassa reserve is with 42,400 km2 the largest conservation area of the country, containing also the greatest concentration of wildlife of the country. the two reserved are connected by the Selous – Niassa wildlife corridor, an area of approximately 9000 km2 located entirely on the Tanzanian side which represents an important biological link between the two reserves and consequently for the miombo woodland eco-system. the Selous - Niassa ecosystem is one of the largest trans-boundary natural dry forest eco-regions in Africa. It constitutes one of the largest elephant ranges in the world and contains half of the world remaining wild dog population. it enables migration of wildlife and gene flow and contributing to the conservation of biodiversity. The dataset is organised in tiles, thus there is one folder for each of the Sentinel-2 tiles used. For each of these folders, there are the labels (both in vector and raster format) and the acquisitions organised in different acquisitions times. Specifically, the dataset has the following folder structure: <ul> <li>LandCover_Train_v2: directory for the training dataset of UC2 <ul> <li>00XXX (e.g.: 37MCN): folders with Sentinel-2 tile name. <ul> <li>labels: folder with raster and vector labels. The name of the label files in both format is the name of the Sentinel-2 tile to which they refer to (e.g.: 37MCN.tif/37MCN.shp) <ul> <li>raster: labels in raster format (.tif)</li> <li>vector: labels in vector format (.shp)</li> </ul> </li> <li>AAAAMMDD (e.g.: 20170630): folders with datetime of the Sentinel-2 acquisition. It contains all the 13 bands of the acquisition plus the TCI (True Color Image) in jp2 format.</li> </ul> </li> </ul> </li> <li>LandCover_Test: directory for the test dataset of UC2 <ul> <li>00XXX (e.g.: 37MCN): folders with Sentinel-2 tile name. <ul> <li>labels: folder with raster and vector labels. The name of the label files in both format is the name of the Sentinel-2 tile to which they refer to (e.g.: 37MCN.tif/37MCN.shp) <ul> <li>raster: labels in raster format (.tif)</li> <li>vector: labels in vector format (.shp)</li> </ul> </li> <li>AAAAMMDD (e.g.: 20170630): folders with datetime of the Sentinel-2 acquisition. It contains all the 13 bands of the acquisition plus the TCI (True Color Image) in jp2 format.</li> </ul> </li> </ul> </li> <li>GT_legend.xlsx: excel file with the association between the classes and the assigned number value in the raster version of the labels</li> </ul>

opencc-by-4.0May 2023View details →

zenodo32/100

Best Learned Models: End-to-end Learning for Land Cover Classification using Irregular and Unaligned SITS by Combining Attention-Based Interpolation with Sparse Variational Gaussian Processes

Best learned model learned with the dataset available <a href="http://https://doi.org/10.5281/zenodo.8033058">here</a> for the mTAN-GP, mTAN-MLP, mTAN-LTAE, and raw-LTAE. For further details see the pre-print article "End-to-end Learning for Land Cover Classification using Irregular and Unaligned SITS by Combining Attention-Based Interpolation with Sparse Variational Gaussian Processes ". This article is available : <a href="https://hal.science/hal-04112115">here</a>. The implementation of the models is available in the <a href="https://gitlab.cesbio.omp.eu/belletv/land_cover_southfrance_mtan_gp_irregular_sits">open source repository</a>.

opencc-by-4.0Jun 2023View details →

zenodo32/100

Classification Data set: End-to-end Learning for Land Cover Classification using Irregular and Unaligned SITS by Combining Attention-Based Interpolation with Sparse Variational Gaussian Processes

Classification data set (train, validation, test) from the study area based on 27 tiles on the south of the France. This dataset contains irregular and unaligned SITS with their corresponding masks. 9 different random sampling are provided. This data set was used to train mTAN-GP, mTAN-MLP, mTAN-LTAE and raw-LTAE. For further details see the pre-print article "End-to-end Learning for Land Cover Classification using Irregular and Unaligned SITS by Combining Attention-Based Interpolation with Sparse Variational Gaussian Processes ". This article is available : <a href="https://hal.science/hal-04112115">here</a>. The implementation of the models is available in the <a href="https://gitlab.cesbio.omp.eu/belletv/land_cover_southfrance_mtan_gp_irregular_sits">open source repository</a>.

opencc-by-4.0Jun 2023View details →

zenodo32/100

SyntheWorld: A Large-Scale Synthetic Dataset for Land Cover Mapping and Building Change Detection

Paper Accept by WACV 2024 [paper, supp] [<a href="https://arxiv.org/abs/2309.01907">arXiv</a>] Overview Synthetic datasets, recognized for their cost effectiveness, play a pivotal role in advancing computer vision tasks and techniques. However, when it comes to remote sensing image processing, the creation of synthetic datasets becomes challenging due to the demand for larger-scale and more diverse 3D models. This complexity is compounded by the difficulties associated with real remote sensing datasets, including limited data acquisition and high annotation costs, which amplifies the need for high-quality synthetic alternatives. To address this, we present SyntheWorld, a synthetic dataset unparalleled in quality, diversity, and scale. It includes 40,000 images with submeter-level pixels and fine-grained land cover annotations of eight categories, and it also provides 40,000 pairs of bitemporal image pairs with building change annotations for building change detection task. We conduct experiments on multiple benchmark remote sensing datasets to verify the effectiveness of SyntheWorld and to investigate the conditions under which our synthetic data yield advantages. <pre>Description ------------ This dataset has been designed for land cover mapping and building change detection tasks. File Structure and Content: --------------------------- 1. **1024.zip**: - Contains images of size 1024x1024 with a GSD (Ground Sampling Distance) of 0.6-1m. - `images` and `ss_mask` folders: Used for the land cover mapping task. - `images` folder: Post-event images for building change detection. - `small-pre-images`: Images with a minor off-nadir angle difference compared to post-event images. - `big-pre-images`: Images with a large off-nadir angle difference compared to post-event images. - `cd_mask`: Ground truth for the building change detection task. 2. **512-1.zip**, **512-2.zip**, **512-3.zip**: - Contains images of size 512x512 with a GSD of 0.3-0.6m. - `images` and `ss_mask` folders: Used for the land cover mapping task. - `images` folder: Post-event images for building change detection. - `pre-event` folder: Images for the pre-event phase. - `cd-mask`: Ground truth for building change detection. Land Cover Mapping Class Grep Map: ---------------------------------- class_grey = { "Bareland": 1, "Rangeland": 2, "Developed Space": 3, "Road": 4, "Tree": 5, "Water": 6, "Agriculture land": 7, "Building": 8, }</pre> Reference @misc{song2023syntheworld,       title={SyntheWorld: A Large-Scale Synthetic Dataset for Land Cover Mapping and Building Change Detection},        author={Jian Song and Hongruixuan Chen and Naoto Yokoya},       year={2023},       eprint={2309.01907},       archivePrefix={arXiv},       primaryClass={cs.CV} }

opencc-by-4.0Sep 2023View details →

dryad32/100