{"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/multi-temporal-land-cover-classification-with","title":"Multi-Temporal Land Cover Classification with Sequential Recurrent Encoders","arxiv_id":"1802.02080","date":"2018-02-06","proceeding":"International Journal of Geo-Information 2018 3","authors":["Marc Rußwurm","Marco Körner"],"abstract":"Earth observation (EO) sensors deliver data with daily or weekly temporal\nresolution. Most land use and land cover (LULC) approaches, however, expect\ncloud-free and mono-temporal observations. The increasing temporal capabilities\nof today's sensors enables the use of temporal, along with spectral and spatial\nfeatures. Domains, such as speech recognition or neural machine translation,\nwork with inherently temporal data and, today, achieve impressive results using\nsequential encoder-decoder structures. Inspired by these sequence-to-sequence\nmodels, we adapt an encoder structure with convolutional recurrent layers in\norder to approximate a phenological model for vegetation classes based on a\ntemporal sequence of Sentinel 2 (S2) images. In our experiments, we visualize\ninternal activations over a sequence of cloudy and non-cloudy images and find\nseveral recurrent cells, which reduce the input activity for cloudy\nobservations. Hence, we assume that our network has learned cloud-filtering\nschemes solely from input data, which could alleviate the need for tedious\ncloud-filtering as a preprocessing step for many EO approaches. Moreover, using\nunfiltered temporal series of top-of-atmosphere (TOA) reflectance data, we\nachieved in our experiments state-of-the-art classification accuracies on a\nlarge number of crop classes with minimal preprocessing compared to other\nclassification approaches.","url_abs":"http://arxiv.org/abs/1802.02080v4","url_pdf":"http://arxiv.org/pdf/1802.02080v4.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":[],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"earth-observation","task_name":"Earth Observation"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"land-cover-classification","task_name":"Land Cover Classification"},{"task_slug":"machine-translation","task_name":"Machine Translation"},{"task_slug":"speech-recognition","task_name":"Speech Recognition"},{"task_slug":"unet-segmentation","task_name":"UNET Segmentation"},{"task_slug":"speech-recognition-1","task_name":"speech-recognition"}],"methods":[],"datasets_introduced":[{"slug":"munich-sentinel2-crop-segmentation","name":"Munich Sentinel2 Crop Segmentation","full_name":""}],"methods_introduced":[],"results":[{"leaderboard":"/sota/unet-segmentation-on-munich-sentinel2-crop-1","task":"UNET Segmentation","dataset":"Munich Sentinel2 Crop Segmentation","model":"Sequential Recurrent Encoders","rank_in_archive_order":4,"of":5,"metrics":{"Overall Accuracy":"89.60"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1802.02080","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}