{"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/semask-semantically-masked-transformers-for-1","title":"SeMask: Semantically Masked Transformers for Semantic Segmentation","arxiv_id":"2112.12782","date":"2021-12-23","proceeding":"arXiv 2021 12","authors":["Jitesh Jain","Anukriti Singh","Nikita Orlov","Zilong Huang","Jiachen Li","Steven Walton","Humphrey Shi"],"abstract":"Finetuning a pretrained backbone in the encoder part of an image transformer network has been the traditional approach for the semantic segmentation task. However, such an approach leaves out the semantic context that an image provides during the encoding stage. This paper argues that incorporating semantic information of the image into pretrained hierarchical transformer-based backbones while finetuning improves the performance considerably. To achieve this, we propose SeMask, a simple and effective framework that incorporates semantic information into the encoder with the help of a semantic attention operation. In addition, we use a lightweight semantic decoder during training to provide supervision to the intermediate semantic prior maps at every stage. Our experiments demonstrate that incorporating semantic priors enhances the performance of the established hierarchical encoders with a slight increase in the number of FLOPs. We provide empirical proof by integrating SeMask into Swin Transformer and Mix Transformer backbones as our encoder paired with different decoders. Our framework achieves a new state-of-the-art of 58.25% mIoU on the ADE20K dataset and improvements of over 3% in the mIoU metric on the Cityscapes dataset. The code and checkpoints are publicly available at https://github.com/Picsart-AI-Research/SeMask-Segmentation .","url_abs":"https://arxiv.org/abs/2112.12782v3","url_pdf":"https://arxiv.org/pdf/2112.12782v3.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":"semask-semantically-masked-transformers-for-1","repo_url":"https://github.com/Picsart-AI-Research/SeMask-Segmentation","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[{"method_slug":"absolute-position-encodings","method_name":"Absolute Position Encodings"},{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"label-smoothing","method_name":"Label Smoothing"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"position-wise-feed-forward-layer","method_name":"Position-Wise Feed-Forward Layer"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"stochastic-depth","method_name":"Stochastic Depth"},{"method_slug":"swin-transformer","method_name":"Swin Transformer"},{"method_slug":"transformer","method_name":"Transformer"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/semantic-segmentation-on-ade20k","task":"Semantic Segmentation","dataset":"ADE20K","model":"SeMask (SeMask Swin-L FaPN-Mask2Former)","rank_in_archive_order":23,"of":235,"metrics":{"Validation mIoU":"58.2"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k","task":"Semantic Segmentation","dataset":"ADE20K","model":"SeMask (SeMask Swin-L MSFaPN-Mask2Former)","rank_in_archive_order":24,"of":235,"metrics":{"Validation mIoU":"58.2"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k","task":"Semantic Segmentation","dataset":"ADE20K","model":"SeMask (SeMask Swin-L Mask2Former)","rank_in_archive_order":32,"of":235,"metrics":{"Validation mIoU":"57.5"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k","task":"Semantic Segmentation","dataset":"ADE20K","model":"SeMask(SeMask Swin-L MSFaPN-Mask2Former, single-scale)","rank_in_archive_order":36,"of":235,"metrics":{"Validation mIoU":"57.0"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k","task":"Semantic Segmentation","dataset":"ADE20K","model":"SeMask (SeMask Swin-L MaskFormer)","rank_in_archive_order":41,"of":235,"metrics":{"Validation mIoU":"56.2"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k","task":"Semantic Segmentation","dataset":"ADE20K","model":"SeMask (SeMask Swin-L FPN)","rank_in_archive_order":79,"of":235,"metrics":{"Validation mIoU":"53.52"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k","task":"Semantic Segmentation","dataset":"ADE20K","model":"SeMask (SeMask Swin-B FPN)","rank_in_archive_order":106,"of":235,"metrics":{"Params (M)":"96","Validation mIoU":"50.98"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k","task":"Semantic Segmentation","dataset":"ADE20K","model":"SeMask (SeMask Swin-S FPN)","rank_in_archive_order":159,"of":235,"metrics":{"Params (M)":"56","Validation mIoU":"47.63"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k","task":"Semantic Segmentation","dataset":"ADE20K","model":"SeMask (SeMask Swin-T FPN)","rank_in_archive_order":211,"of":235,"metrics":{"Params (M)":"35","Validation mIoU":"43.16"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k-val","task":"Semantic Segmentation","dataset":"ADE20K val","model":"SeMask (SeMask Swin-L FaPN-Mask2Former)","rank_in_archive_order":15,"of":95,"metrics":{"mIoU":"58.2"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k-val","task":"Semantic Segmentation","dataset":"ADE20K val","model":"SeMask (SeMask Swin-L MSFaPN-Mask2Former)","rank_in_archive_order":16,"of":95,"metrics":{"mIoU":"58.2"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k-val","task":"Semantic Segmentation","dataset":"ADE20K val","model":"SeMask (SeMask Swin-L Mask2Former)","rank_in_archive_order":20,"of":95,"metrics":{"mIoU":"57.5"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k-val","task":"Semantic Segmentation","dataset":"ADE20K val","model":"SeMask (SeMask Swin-L MSFaPN-Mask2Former, single-scale)","rank_in_archive_order":23,"of":95,"metrics":{"mIoU":"57.0"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k-val","task":"Semantic Segmentation","dataset":"ADE20K val","model":"SeMask (SeMask Swin-L MaskFormer)","rank_in_archive_order":26,"of":95,"metrics":{"mIoU":"56.2"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k-val","task":"Semantic Segmentation","dataset":"ADE20K val","model":"SeMask (SeMask Swin-L FPN)","rank_in_archive_order":40,"of":95,"metrics":{"mIoU":"53.5"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-cityscapes-val","task":"Semantic Segmentation","dataset":"Cityscapes val","model":"SeMask (SeMask Swin-L Mask2Former)","rank_in_archive_order":11,"of":99,"metrics":{"mIoU":"84.98"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-cityscapes-val","task":"Semantic Segmentation","dataset":"Cityscapes val","model":"SeMask (SeMask Swin-L FPN)","rank_in_archive_order":51,"of":99,"metrics":{"mIoU":"80.39"},"uses_additional_data":true}],"syntology":{"syntology_url":"https://syntology.ai/paper/2112.12782","atlas_url":"https://app.syntology.ai/?focus=2112.12782","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}