Dataset for Land/Water Semantic Segmentation in Tonga and other Pacific regions
Published: 24 February 2025| Version 2 | DOI: 10.17632/mfc95sgrbf.2
Contributor:
Riccardo PercacciDescription
Dataset for training CNN models (e.g. U-Net) for land/water semantic segmentation of Sentinel-2 images. Features 424 Sentinel-2 images of size 256x256 pixels, with corresponding targets (binary segmentation maps). Weight maps are included for training using a weighted loss function, in order to improve segmentation on important features, specifically small volcanic islands. The weights of the pre-trained models used for our research are included as .keras files. Code snippets include essential functionality to load the models, and pre-process Earth Engine images for input.
Files
Institutions
Universita degli Studi di Trieste Dipartimento di Matematica e Geoscienze, Universita degli Studi di Trieste
Categories
Use of Computers in Earth Sciences, Remote Sensing, Image Segmentation, Seamount, Tonga, Volcanism, U-Net