{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/temporal-convolutional-neural-network-for-the","title":"Temporal Convolutional Neural Network for the Classification of Satellite Image Time Series","arxiv_id":"1811.10166","date":"2018-11-26","proceeding":null,"authors":["Charlotte Pelletier","Geoffrey I. Webb","Francois Petitjean"],"abstract":"New remote sensing sensors now acquire high spatial and spectral Satellite\nImage Time Series (SITS) of the world. These series of images are a key\ncomponent of classification systems that aim at obtaining up-to-date and\naccurate land cover maps of the Earth's surfaces. More specifically, the\ncombination of the temporal, spectral and spatial resolutions of new SITS makes\npossible to monitor vegetation dynamics. Although traditional classification\nalgorithms, such as Random Forest (RF), have been successfully applied for SITS\nclassification, these algorithms do not make the most of the temporal domain.\nConversely, some approaches that take into account the temporal dimension have\nrecently been tested, especially Recurrent Neural Networks (RNNs). This paper\nproposes an exhaustive study of another deep learning approaches, namely\nTemporal Convolutional Neural Networks (TempCNNs) where convolutions are\napplied in the temporal dimension. The goal is to quantitatively and\nqualitatively evaluate the contribution of TempCNNs for SITS classification.\nThis paper proposes a set of experiments performed on one million time series\nextracted from 46 Formosat-2 images. The experimental results show that\nTempCNNs are more accurate than RF and RNNs, that are the current state of the\nart for SITS classification. We also highlight some differences with results\nobtained in computer vision, e.g. about pooling layers. Moreover, we provide\nsome general guidelines on the network architecture, common regularization\nmechanisms, and hyper-parameter values such as batch size. Finally, we assess\nthe visual quality of the land cover maps produced by TempCNNs.","url_abs":"http://arxiv.org/abs/1811.10166v2","url_pdf":"http://arxiv.org/pdf/1811.10166v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"temporal-convolutional-neural-network-for-the","repo_url":"https://github.com/charlotte-pel/temporalCNN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"GPL-3.0"}}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"time-series-1","task_name":"Time Series"},{"task_slug":"time-series","task_name":"Time Series Analysis"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1811.10166","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}