CatLC: Catalonia Multiresolution Land Cover Dataset
Keywords: remote sensing, land cover, dataset, segmentation
Abstract: Traditional natural image datasets are very rich. However, only a few remote sensing datasets are available and cover a tiny territory or cover a larger one with low spatial resolution and/or few classes. In this paper, we present the Catalonia Multiresolution Land Cover Dataset (CatLC), a remote sensing dataset. The dataset contains images at different spatial resolutions captured by both aircraft and satellites (Sentinel-1 and Sentinel-2), in addition to topographic maps. All this dataset has been created with images from the Cartographic and Geological Institute of Catalonia (ICGC) catalogs and the European Space Agency (ESA). The ICGC's land cover ground truth accompanies these images with 41 classes at a spatial resolution of 1 m in an area of 32000 km2, covering the Spanish region of Catalonia. CatLC is a multilayer, multiresolution, multimodal, multitemporal dataset, which has excellent potential for the Artificial Intelligence (AI) community and the exploration of modeling methodologies. Land cover maps are used in different realms such as forestry for inventory area estimates, hydrology regarding microclimatic variables, agriculture to improve irrigation or geology in geohazards, and risk identification and assessment. Therefore, accurate and updated knowledge about land changes is essential for territory management with different purposes over multiple fields. Using various combinations of the images from the dataset, we offer a benchmark that could serve as a starting point to explore artificial intelligence techniques for remote sensing segmentation purposes. In this vein, CatLC dataset aims to engage with computer vision experts interested in remote sensing and stimulate research and development.
URL: https://www.icgc.cat/en/Downloads/Aerial-and-satellite-images/Contingut/Catalonia-Multi-resolution-Landcover-Dataset-CatLC
12 Replies
Loading