{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,5,1]],"date-time":"2026-05-01T16:59:42Z","timestamp":1777654782726,"version":"3.51.4"},"reference-count":81,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2024,6,18]],"date-time":"2024-06-18T00:00:00Z","timestamp":1718668800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Big Data"],"abstract":"<jats:p>Semantic segmentation models trained on annotated data fail to generalize well when the input data distribution changes over extended time period, leading to requiring re-training to maintain performance. Classic unsupervised domain adaptation (UDA) attempts to address a similar problem when there is target domain with no annotated data points through transferring knowledge from a source domain with annotated data. We develop an online UDA algorithm for semantic segmentation of images that improves model generalization on unannotated domains in scenarios where source data access is restricted during adaptation. We perform model adaptation by minimizing the distributional distance between the source latent features and the target features in a shared embedding space. Our solution promotes a shared domain-agnostic latent feature space between the two domains, which allows for classifier generalization on the target dataset. To alleviate the need of access to source samples during adaptation, we approximate the source latent feature distribution via an appropriate surrogate distribution, in this case a Gaussian mixture model (GMM).<\/jats:p>","DOI":"10.3389\/fdata.2024.1359317","type":"journal-article","created":{"date-parts":[[2024,6,18]],"date-time":"2024-06-18T05:11:44Z","timestamp":1718687504000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":4,"title":["Source-free domain adaptation for semantic image segmentation using internal representations"],"prefix":"10.3389","volume":"7","author":[{"given":"Serban","family":"Stan","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Mohammad","family":"Rostami","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1965","published-online":{"date-parts":[[2024,6,18]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1701.07875","article-title":"Wasserstein GAN","author":"Arjovsky","year":"2017","journal-title":"arXiv"},{"key":"B2","doi-asserted-by":"publisher","first-page":"1369","DOI":"10.3390\/rs11111369","article-title":"Unsupervised domain adaptation using generative adversarial networks for semantic segmentation of aerial images","volume":"11","author":"Benjdira","year":"2019","journal-title":"Remote Sens"},{"key":"B3","first-page":"447","article-title":"\u201cDeepjdot: deep joint distribution optimal transport for unsupervised domain adaptation,\u201d","volume-title":"Proceedings of the European Conference on Computer Vision (ECCV)","author":"Bhushan Damodaran","year":"2018"},{"key":"B4","doi-asserted-by":"publisher","first-page":"223","DOI":"10.1137\/16M1080173","article-title":"Optimization methods for large-scale machine learning","volume":"60","author":"Bottou","year":"2018","journal-title":"Siam Rev"},{"key":"B5","first-page":"3722","article-title":"\u201cUnsupervised pixel-level domain adaptation with generative adversarial networks,\u201d","volume-title":"Proceedings of the IEEE conference on computer vision and pattern recognition","author":"Bousmalis","year":"2017"},{"key":"B6","first-page":"43","article-title":"\u201cSemantic-aware generative adversarial nets for unsupervised domain adaptation in chest X-ray segmentation,\u201d","volume-title":"International workshop on machine learning in medical imaging","author":"Chen","year":""},{"key":"B7","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1706.05587","article-title":"Rethinking atrous convolution for semantic image segmentation","author":"Chen","year":"","journal-title":"arXiv"},{"key":"B8","doi-asserted-by":"crossref","DOI":"10.1007\/978-3-030-01234-2_49","article-title":"\u201cEncoder-decoder with atrous separable convolution for semantic image segmentation,\u201d","volume-title":"ECCV","author":"Chen","year":""},{"key":"B9","first-page":"1841","article-title":"\u201cLearning semantic segmentation from synthetic data: a geometrically guided input-output adaptation approach,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Chen","year":"2019"},{"key":"B10","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1704.08509","article-title":"No more discrimination: cross city adaptation of road scene segmenters","author":"Chen","year":"","journal-title":"arXiv"},{"key":"B11","first-page":"6830","article-title":"\u201cSelf-ensembling with gan-based data augmentation for domain adaptation in semantic segmentation,\u201d","volume-title":"Proceedings of the IEEE\/CVF International Conference on Computer Vision","author":"Choi","year":"2019"},{"key":"B12","first-page":"3213","article-title":"\u201cThe cityscapes dataset for semantic urban scene understanding,\u201d","volume-title":"Proceedings of the IEEE conference on computer vision and pattern recognition","author":"Cordts","year":"2016"},{"key":"B13","first-page":"689","article-title":"\u201cOnline methods for multi-domain learning and adaptation,\u201d","volume-title":"Proceedings of the Conference on Empirical Methods in Natural Language Processing","author":"Dredze","year":"2008"},{"key":"B14","doi-asserted-by":"publisher","first-page":"1341","DOI":"10.1109\/TITS.2020.2972974","article-title":"Deep multi-modal object detection and semantic segmentation for autonomous driving: datasets, methods, and challenges","volume":"22","author":"Feng","year":"2020","journal-title":"IEEE Trans. Intell. Transp. Syst"},{"key":"B15","doi-asserted-by":"crossref","first-page":"352","DOI":"10.1109\/ALLERTON.2019.8919960","article-title":"\u201cLearning a domain-invariant embedding for unsupervised domain adaptation using class-conditioned distribution alignment,\u201d","volume-title":"2019 57th Annual Allerton Conference on Communication, Control, and Computing (Allerton)","author":"Gabourie","year":"2019"},{"key":"B16","unstructured":"GoodfellowI. BengioY. CourvilleA. Cambridge, MAMIT PressDeep Learning2016"},{"key":"B17","first-page":"2672","article-title":"\u201cGenerative adversarial nets,\u201d","volume-title":"Advances in Neural Information Processing Systems","author":"Goodfellow","year":"2014"},{"key":"B18","doi-asserted-by":"publisher","first-page":"139","DOI":"10.1145\/3422622","article-title":"Generative adversarial networks","volume":"63","author":"Goodfellow","year":"2020","journal-title":"Commun. ACM"},{"key":"B19","doi-asserted-by":"publisher","first-page":"1173","DOI":"10.1109\/TBME.2021.3117407","article-title":"Domain adaptation for medical image analysis: a survey","volume":"69","author":"Guan","year":"2021","journal-title":"IEEE Trans. Biomed. Eng"},{"key":"B20","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1512.03385","article-title":"Deep residual learning for image recognition","author":"He","year":"2015","journal-title":"arXiv"},{"key":"B21","first-page":"1989","article-title":"\u201cCycada: cycle-consistent adversarial domain adaptation,\u201d","volume-title":"International conference on machine learning","author":"Hoffman","year":""},{"key":"B22","first-page":"1989","article-title":"\u201cCyCADA: cycle-consistent adversarial domain adaptation,\u201d","volume-title":"International Conference on Machine Learning","author":"Hoffman","year":""},{"key":"B23","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1612.02649","article-title":"FCNS in the wild: pixel-level adversarial and constraint-based adaptation","author":"Hoffman","year":"2016","journal-title":"arXiv"},{"key":"B24","first-page":"1335","article-title":"\u201cConditional generative adversarial network for structured domain adaptation,\u201d","volume-title":"Proceedings of the IEEE conference on computer vision and pattern recognition","author":"Hong","year":"2018"},{"key":"B25","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1802.07934","article-title":"Adversarial learning for semi-supervised semantic segmentation","author":"Hung","year":"2018","journal-title":"arXiv"},{"key":"B26","first-page":"577","article-title":"\u201cOnline domain adaptation of a pre-trained cascade of classifiers,\u201d","volume-title":"Proceedings of the 2011 IEEE Conference on Computer Vision and Pattern Recognition","author":"Jain","year":"2011"},{"key":"B27","doi-asserted-by":"publisher","first-page":"32650","DOI":"10.1109\/ACCESS.2021.3060821","article-title":"Analysis based on recent deep learning approaches applied in real-time multi-object tracking: a review","volume":"9","author":"Kalake","year":"2021","journal-title":"IEEE Access"},{"key":"B28","doi-asserted-by":"publisher","first-page":"508","DOI":"10.1109\/TAI.2021.3110179","article-title":"Domain adaptation without source data","volume":"2","author":"Kim","year":"2021","journal-title":"IEEE Trans. Artif. Intell"},{"key":"B29","first-page":"32","article-title":"\u201cGeneralized sliced wasserstein distances,\u201d","volume-title":"Advances in neural information processing systems","author":"Kolouri","year":"2019"},{"key":"B30","first-page":"5258","article-title":"\u201cSliced wasserstein kernels for probability distributions,\u201d","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Kolouri","year":"2016"},{"key":"B31","first-page":"7046","article-title":"\u201cGeneralize then adapt: source-free domain adaptive semantic segmentation,\u201d","volume-title":"Proceedings of the IEEE\/CVF International Conference on Computer Vision (ICCV)","author":"Kundu","year":""},{"key":"B32","article-title":"\u201cUniversal source-free domain adaptation,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","author":"Kundu","year":""},{"key":"B33","first-page":"3361","article-title":"\u201cConvolutional networks for images, speech, and time series,\u201d","author":"LeCun","year":"1995","journal-title":"The Handbook of Brain Theory and Neural Networks"},{"key":"B34","first-page":"10285","article-title":"\u201cSliced wasserstein discrepancy for unsupervised domain adaptation,\u201d","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Lee","year":"2019"},{"key":"B35","first-page":"13936","article-title":"\u201cEnhanced transport distance for unsupervised domain adaptation,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Li","year":""},{"key":"B36","doi-asserted-by":"publisher","first-page":"101765","DOI":"10.1016\/j.media.2020.101765","article-title":"Multi-site fmri analysis using privacy-preserving federated learning and domain adaptation: abide results","volume":"65","author":"Li","year":"","journal-title":"Med. Image Anal"},{"key":"B37","doi-asserted-by":"crossref","DOI":"10.1109\/CVPR.2017.106","article-title":"\u201cFeature pyramid networks for object detection,\u201d","volume-title":"2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)","author":"Lin","year":"2017"},{"key":"B38","first-page":"1215","article-title":"\u201cSource-free domain adaptation for semantic segmentation,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Liu","year":"2021"},{"key":"B39","first-page":"3431","article-title":"\u201cFully convolutional networks for semantic segmentation,\u201d","volume-title":"Proceedings of the IEEE conference on computer vision and pattern recognition","author":"Long","year":"2015"},{"key":"B40","article-title":"\u201cSemantic segmentation using adversarial networks,\u201d","volume-title":"NIPS Workshop on Adversarial Training","author":"Luc","year":"2016"},{"key":"B41","doi-asserted-by":"publisher","first-page":"861","DOI":"10.21105\/joss.00861","article-title":"UMAP: uniform manifold approximation and projection","volume":"3","author":"McInnes","year":"2018","journal-title":"J. Open Source Softw"},{"key":"B42","article-title":"\u201cImage to image translation for domain adaptation,\u201d","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Murez","year":"2017"},{"key":"B43","doi-asserted-by":"crossref","first-page":"737","DOI":"10.1007\/978-3-319-71246-8_45","article-title":"\u201cTheoretical analysis of domain adaptation with optimal transport,\u201d","volume-title":"Joint European Conference on Machine Learning and Knowledge Discovery in Databases","author":"Redko","year":"2017"},{"key":"B44","first-page":"102","article-title":"\u201cPlaying for data: ground truth from computer games,\u201d","volume-title":"European conference on computer vision","author":"Richter","year":"2016"},{"key":"B45","doi-asserted-by":"crossref","first-page":"1312","DOI":"10.1109\/IVS.2019.8813888","article-title":"\u201cBridging the day and night domain gap for semantic segmentation,\u201d","volume-title":"2019 IEEE Intelligent Vehicles Symposium (IV)","author":"Romera","year":"2019"},{"key":"B46","doi-asserted-by":"publisher","DOI":"10.1007\/978-3-319-24574-4_28","article-title":"\u201cU-net: convolutional networks for biomedical image segmentation,\u201d","author":"Ronneberger","year":"2015","journal-title":"Medical Image Computing and Computer-Assisted Intervention"},{"key":"B47","first-page":"3234","article-title":"\u201cThe synthia dataset: a large collection of synthetic images for semantic segmentation of urban scenes,\u201d","volume-title":"Proceedings of the IEEE conference on computer vision and pattern recognition","author":"Ros","year":"2016"},{"key":"B48","author":"Rostami","year":"2019","journal-title":"Learning Transferable Knowledge Through Embedding Spaces"},{"key":"B49","doi-asserted-by":"publisher","first-page":"9623","DOI":"10.1609\/aaai.v37i8.26151","article-title":"Overcoming concept shift in domain-aware settings through consolidated internal distributions","volume":"37","author":"Rostami","year":"2023","journal-title":"Proc. AAAI Conf. Artif. Intell"},{"key":"B50","doi-asserted-by":"publisher","DOI":"10.3390\/rs11111374","article-title":"Deep transfer learning for few-shot sar image classification","author":"Rostami","year":"2019","journal-title":"Remote Sens"},{"key":"B51","first-page":"3723","article-title":"\u201cMaximum classifier discrepancy for unsupervised domain adaptation,\u201d","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Saito","year":"2018"},{"key":"B52","first-page":"687","article-title":"\u201cModel adaptation with synthetic and real data for semantic dense foggy scene understanding,\u201d","volume-title":"Proceedings of the European Conference on Computer Vision (ECCV)","author":"Sakaridis","year":"2018"},{"key":"B53","doi-asserted-by":"crossref","first-page":"771","DOI":"10.1109\/3DV50981.2020.00087","article-title":"\u201cSF-UDA 3D: source-free unsupervised domain adaptation for lidar-based 3D object detection,\u201d","volume-title":"2020 International Conference on 3D Vision (3DV)","author":"Saltori","year":"2020"},{"key":"B54","first-page":"3752","article-title":"\u201cLearning from synthetic data: addressing domain shift for semantic segmentation,\u201d","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Sankaranarayanan","year":"2018"},{"key":"B55","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1409.1556","author":"Simonyan","year":"2014","journal-title":"Very deep convolutional networks for large-scale image recognition. arXiv"},{"key":"B56","doi-asserted-by":"publisher","first-page":"66","DOI":"10.1145\/2766963","article-title":"Convolutional wasserstein distances: efficient optimal transportation on geometric domains","volume":"34","author":"Solomon","year":"2015","journal-title":"ACM Trans. Graph"},{"key":"B57","doi-asserted-by":"publisher","first-page":"2593","DOI":"10.1609\/aaai.v35i3.16362","article-title":"Unsupervised model adaptation for continual semantic segmentation","volume":"35","author":"Stan","year":"2021","journal-title":"Proc. AAAI Conf. Artif. Intell"},{"key":"B58","doi-asserted-by":"crossref","DOI":"10.1109\/ICCV48922.2021.00717","article-title":"\u201cSegmenter: transformer for semantic segmentation,\u201d","volume-title":"2021 IEEE\/CVF International Conference on Computer Vision (ICCV)","author":"Strudel","year":"2021"},{"key":"B59","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1109\/ECMR50962.2021.9568802","article-title":"\u201cAerial-pass: panoramic annular scene segmentation in drone videos,\u201d","volume-title":"2021 European Conference on Mobile Robots (ECMR)","author":"Sun","year":"2021"},{"key":"B60","article-title":"Hierarchical multi-scale attention for semantic segmentation","author":"Tao","year":"2020","journal-title":"Arxiv."},{"key":"B61","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1711.01558","article-title":"Wasserstein auto-encoders","author":"Tolstikhin","year":"2017","journal-title":"arXiv"},{"key":"B62","first-page":"7472","article-title":"\u201cLearning to adapt structured output space for semantic segmentation,\u201d","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Tsai","year":"2018"},{"key":"B63","first-page":"2517","article-title":"\u201cAdvent: adversarial entropy minimization for domain adaptation in semantic segmentation,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Vu","year":"2019"},{"key":"B64","doi-asserted-by":"crossref","DOI":"10.1007\/978-3-030-58548-8_7","article-title":"\u201cAxial-deeplab: stand-alone axial-attention for panoptic segmentation,\u201d","volume-title":"European Conference on Computer Vision (ECCV)","author":"Wang","year":"2020"},{"key":"B65","doi-asserted-by":"publisher","first-page":"135","DOI":"10.1016\/j.neucom.2018.05.083","article-title":"Deep visual domain adaptation: a survey","volume":"312","author":"Wang","year":"2018","journal-title":"Neurocomputing"},{"key":"B66","doi-asserted-by":"publisher","first-page":"4376","DOI":"10.1109\/TIP.2019.2910667","article-title":"Weakly supervised adversarial domain adaptation for semantic segmentation in urban scenes","volume":"28","author":"Wang","year":"2019","journal-title":"IEEE Trans. Image Process"},{"key":"B67","doi-asserted-by":"crossref","DOI":"10.1109\/CVPR.2018.00759","article-title":"\u201cRevisiting dilated convolution: a simple approach for weakly- and semi-supervised semantic segmentation,\u201d","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)","author":"Wei","year":"2018"},{"key":"B68","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1145\/3400066","article-title":"A survey of unsupervised deep domain adaptation","volume":"11","author":"Wilson","year":"2020","journal-title":"ACM Trans. Intell. Syst. Technol."},{"key":"B69","doi-asserted-by":"publisher","first-page":"550","DOI":"10.1109\/THMS.2016.2608931","article-title":"Online and offline domain adaptation for reducing bci calibration effort","volume":"47","author":"Wu","year":"2016","journal-title":"EEE Trans. Hum. Mach. Syst"},{"key":"B70","first-page":"518","article-title":"\u201cDcan: dual channel-wise alignment networks for unsupervised scene adaptation,\u201d","volume-title":"Proceedings of the European Conference on Computer Vision (ECCV)","author":"Wu","year":"2018"},{"key":"B71","doi-asserted-by":"publisher","first-page":"32","DOI":"10.1109\/MWC.001.2000252","article-title":"Federated-learning-based client scheduling for low-latency wireless communications","volume":"28","author":"Xia","year":"2021","journal-title":"IEEE Wirel. Commun"},{"key":"B72","first-page":"4394","article-title":"\u201cReliable weighted optimal transport for unsupervised domain adaptation,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Xu","year":"2020"},{"key":"B73","first-page":"8978","article-title":"\u201cGeneralized source-free domain adaptation,\u201d","volume-title":"Proceedings of the IEEE\/CVF International Conference on Computer Vision (ICCV)","author":"Yang","year":"2021"},{"key":"B74","first-page":"9011","article-title":"\u201cPhase consistent ecological domain adaptation,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Yang","year":"2020"},{"key":"B75","first-page":"4085","article-title":"\u201cFDA: Fourier domain adaptation for semantic segmentation,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Yang","year":"2020"},{"key":"B76","doi-asserted-by":"crossref","first-page":"3293","DOI":"10.1145\/3474085.3475482","article-title":"\u201cDomain adaptive semantic segmentation without source data,\u201d","volume-title":"Proceedings of the 29th ACM International Conference on Multimedia, MM '21","author":"You","year":"2021"},{"key":"B77","first-page":"2515","article-title":"\u201cDivergence optimization for noisy universal domain adaptation,\u201d","volume-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Yu","year":"2021"},{"key":"B78","first-page":"435","article-title":"\u201cCategory anchor-guided unsupervised domain adaptation for semantic segmentation,\u201d","volume-title":"Advances in Neural Information Processing Systems","author":"Zhang","year":"2019"},{"key":"B79","first-page":"2020","article-title":"\u201cCurriculum domain adaptation for semantic segmentation of urban scenes,\u201d","volume-title":"Proceedings of the IEEE International Conference on Computer Vision","author":"Zhang","year":"2017"},{"key":"B80","first-page":"669","article-title":"\u201cInstance-level segmentation for autonomous driving with deep densely connected MRFS,\u201d","volume-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition","author":"Zhang","year":"2016"},{"key":"B81","first-page":"2223","article-title":"\u201cUnpaired image-to-image translation using cycle-consistent adversarial networks,\u201d","volume-title":"ICCV","author":"Zhu","year":"2017"}],"container-title":["Frontiers in Big Data"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fdata.2024.1359317\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2024,11,22]],"date-time":"2024-11-22T02:12:20Z","timestamp":1732241540000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fdata.2024.1359317\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,6,18]]},"references-count":81,"alternative-id":["10.3389\/fdata.2024.1359317"],"URL":"https:\/\/doi.org\/10.3389\/fdata.2024.1359317","relation":{},"ISSN":["2624-909X"],"issn-type":[{"value":"2624-909X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2024,6,18]]},"article-number":"1359317"}}