{"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/ternausnetv2-fully-convolutional-network-for","title":"TernausNetV2: Fully Convolutional Network for Instance Segmentation","arxiv_id":"1806.00844","date":"2018-06-03","proceeding":null,"authors":["Vladimir I. Iglovikov","Selim Seferbekov","Alexander V. Buslaev","Alexey Shvets"],"abstract":"The most common approaches to instance segmentation are complex and use\ntwo-stage networks with object proposals, conditional random-fields, template\nmatching or recurrent neural networks. In this work we present TernausNetV2 - a\nsimple fully convolutional network that allows extracting objects from a\nhigh-resolution satellite imagery on an instance level. The network has popular\nencoder-decoder type of architecture with skip connections but has a few\nessential modifications that allows using for semantic as well as for instance\nsegmentation tasks. This approach is universal and allows to extend any network\nthat has been successfully applied for semantic segmentation to perform\ninstance segmentation task. In addition, we generalize network encoder that was\npre-trained for RGB images to use additional input channels. It makes possible\nto use transfer learning from visual to a wider spectral range. For\nDeepGlobe-CVPR 2018 building detection sub-challenge, based on public\nleaderboard score, our approach shows superior performance in comparison to\nother methods. The source code corresponding pre-trained weights are publicly\navailable at https://github.com/ternaus/TernausNetV2","url_abs":"http://arxiv.org/abs/1806.00844v2","url_pdf":"http://arxiv.org/pdf/1806.00844v2.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":"ternausnetv2-fully-convolutional-network-for","repo_url":"https://github.com/ternaus/TernausNetV2","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null},{"paper_slug":"ternausnetv2-fully-convolutional-network-for","repo_url":"https://github.com/minerva-ml/open-solution-mapping-challenge","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"ternausnetv2-fully-convolutional-network-for","repo_url":"https://github.com/neptune-ai/open-solution-mapping-challenge","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"instance-segmentation","task_name":"Instance Segmentation"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"},{"task_slug":"template-matching","task_name":"Template Matching"},{"task_slug":"transfer-learning","task_name":"Transfer Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}