{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,6,19]],"date-time":"2026-06-19T16:23:05Z","timestamp":1781886185034,"version":"3.54.5"},"reference-count":29,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2025,1,9]],"date-time":"2025-01-09T00:00:00Z","timestamp":1736380800000},"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. Artif. Intell."],"abstract":"<jats:sec><jats:title>Introduction<\/jats:title><jats:p>Active learning can significantly decrease the labeling cost of deep learning workflows by prioritizing the limited labeling budget to high-impact data points that have the highest positive impact on model accuracy. Active learning is especially useful for semantic segmentation tasks where we can selectively label only a few high-impact regions within these high-impact images. Most established regional active learning algorithms deploy a static-budget querying strategy where a fixed percentage of regions are queried in each image. A static budget could result in over- or under-labeling images as the number of high-impact regions in each image can vary.<\/jats:p><\/jats:sec><jats:sec><jats:title>Methods<\/jats:title><jats:p>In this paper, we present a novel dynamic-budget superpixel querying strategy that can query the optimal numbers of high-uncertainty superpixels in an image to improve the querying efficiency of regional active learning algorithms designed for semantic segmentation.<\/jats:p><\/jats:sec><jats:sec><jats:title>Results<\/jats:title><jats:p>For two distinct datasets, we show that by allowing a dynamic budget for each image, the active learning algorithm is more effective compared to static-budget querying at the same low total labeling budget. We investigate both low- and high-budget scenarios and the impact of superpixel size on our dynamic active learning scheme. In a low-budget scenario, our dynamic-budget querying outperforms static-budget querying by 5.6% mIoU on a specialized agriculture field image dataset and 2.4% mIoU on Cityscapes.<\/jats:p><\/jats:sec><jats:sec><jats:title>Discussion<\/jats:title><jats:p>The presented dynamic-budget querying strategy is simple, effective, and can be easily adapted to other regional active learning algorithms to further improve the data efficiency of semantic segmentation tasks.<\/jats:p><\/jats:sec>","DOI":"10.3389\/frai.2024.1498956","type":"journal-article","created":{"date-parts":[[2025,1,9]],"date-time":"2025-01-09T06:13:00Z","timestamp":1736403180000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":2,"title":["Dynamic-budget superpixel active learning for semantic segmentation"],"prefix":"10.3389","volume":"7","author":[{"given":"Yuemin","family":"Wang","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ian","family":"Stavness","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2025,1,9]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"2274","DOI":"10.1109\/TPAMI.2012.120","article-title":"Slic superpixels compared to state-of-the-art superpixel methods","volume":"34","author":"Achanta","year":"2012","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell"},{"key":"B2","doi-asserted-by":"publisher","first-page":"10988","DOI":"10.1109\/CVPR46437.2021.01084","article-title":"\u201cRevisiting superpixels for active learning in semantic segmentation with realistic annotation costs,\u201d","author":"Cai","year":"2021","journal-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition"},{"key":"B3","doi-asserted-by":"publisher","first-page":"1045","DOI":"10.1177\/0278364917720510","article-title":"Agricultural robot dataset for plant classification, localization and mapping on sugar beet fields","volume":"36","author":"Chebrolu","year":"2017","journal-title":"Int. J. Robot. Res"},{"key":"B4","doi-asserted-by":"publisher","first-page":"801","DOI":"10.1007\/978-3-030-01234-2_49","article-title":"\u201cEncoder-decoder with atrous separable convolution for semantic image segmentation,\u201d","author":"Chen","year":"2018","journal-title":"Proceedings of the European Conference on Computer Vision"},{"key":"B5","doi-asserted-by":"publisher","first-page":"1311","DOI":"10.1145\/2505515.2505528","article-title":"\u201cFeedback-driven multiclass active learning for data streams,\u201d","author":"Cheng","year":"2013","journal-title":"Proceedings of the 22nd ACM international conference on Information & Knowledge Management"},{"key":"B6","doi-asserted-by":"publisher","DOI":"10.21236\/ADA295617","article-title":"\u201cActive learning with statistical models,\u201d","author":"Cohn","year":"1994","journal-title":"Advances in Neural Information Processing Systems"},{"key":"B7","doi-asserted-by":"publisher","first-page":"3213","DOI":"10.1109\/CVPR.2016.350","article-title":"\u201cThe cityscapes dataset for semantic urban scene understanding,\u201d","author":"Cordts","year":"2016","journal-title":"Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition"},{"key":"B8","doi-asserted-by":"publisher","first-page":"167","DOI":"10.1023\/B:VISI.0000022288.19776.77","article-title":"Efficient graph-based image segmentation","volume":"59","author":"Felzenszwalb","year":"2004","journal-title":"Int. J. Comput. Vis"},{"key":"B9","article-title":"Bayesian convolutional neural networks with bernoulli approximate variational inference","author":"Gal","year":"2015","journal-title":"arXiv preprint arXiv:1506.02158"},{"key":"B10","first-page":"1050","article-title":"\u201cDropout as a bayesian approximation: representing model uncertainty in deep learning,\u201d","author":"Gal","year":"2016","journal-title":"International Conference on Machine Learning"},{"key":"B11","first-page":"1183","article-title":"\u201cDeep bayesian active learning with image data,\u201d","volume-title":"International Conference on Machine Learning","author":"Gal","year":"2017"},{"key":"B12","doi-asserted-by":"publisher","first-page":"1030","DOI":"10.1109\/TMI.2019.2940555","article-title":"Automated muscle segmentation from clinical ct using bayesian u-net for personalized musculoskeletal modeling","volume":"39","author":"Hiasa","year":"2019","journal-title":"IEEE Trans. Medic. Imag"},{"key":"B13","article-title":"Bayesian active learning for classification and preference learning","author":"Houlsby","year":"2011","journal-title":"arXiv preprint arXiv:1112.5745"},{"key":"B14","first-page":"186","article-title":"The data model concept in statistical mapping","volume":"7","author":"Jenks","year":"1967","journal-title":"Int. Yearb. Cartogr"},{"key":"B15","doi-asserted-by":"crossref","first-page":"b1079","DOI":"10.1109\/AIM.2003.1225492","article-title":"\u201cCrop growth estimation system using machine vision,\u201d","volume-title":"Proceedings 2003 IEEE\/ASME International Conference on Advanced Intelligent Mechatronics (AIM 2003)","author":"Kataoka","year":"2003"},{"key":"B16","first-page":"1","article-title":"Active learning for cost-sensitive classification","volume":"20","author":"Krishnamurthy","year":"2019","journal-title":"J. Mach. Learn. Res"},{"key":"B17","article-title":"Cereals-cost-effective region-based active learning for semantic segmentation","author":"Mackowiak","year":"2018","journal-title":"arXiv preprint arXiv:1810.09726"},{"key":"B18","doi-asserted-by":"crossref","first-page":"996","DOI":"10.1109\/ICPR.2014.181","article-title":"\u201cCompact watershed and preemptive slic: on improving trade-offs of superpixel segmentation algorithms,\u201d","volume-title":"2014 22nd International Conference on Pattern Recognition","author":"Neubert","year":"2014"},{"key":"B19","first-page":"234","article-title":"\u201cU-net: Convolutional networks for biomedical image segmentation,\u201d","volume-title":"Medical Image Computing and Computer-Assisted Intervention\u2013MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, part III 18","author":"Ronneberger","year":"2015"},{"key":"B20","article-title":"Active learning for convolutional neural networks: a core-set approach","author":"Sener","year":"","journal-title":"arXiv preprint arXiv:1708.00489"},{"key":"B21","article-title":"A geometric approach to active learning for convolutional neural networks","author":"Sener","year":"","journal-title":"arXiv preprint arXiv:1708.00489"},{"key":"B22","article-title":"Qualitative analysis of monte carlo dropout","author":"Seoh","year":"2020","journal-title":"arXiv preprint arXiv:2007.01720"},{"key":"B23","author":"Settles","year":"2009","journal-title":"Active learning literature survey"},{"key":"B24","doi-asserted-by":"publisher","first-page":"9433","DOI":"10.1109\/CVPR42600.2020.00945","article-title":"\u201cViewal: active learning with viewpoint entropy for semantic segmentation,\u201d","author":"Siddiqui","year":"2020","journal-title":"Proceedings of the IEEE\/CVF Conference on Computer Vision and Pattern Recognition"},{"key":"B25","doi-asserted-by":"publisher","first-page":"843","DOI":"10.1109\/ICCV.2017.97","article-title":"\u201cRevisiting unreasonable effectiveness of data in deep learning era,\u201d","author":"Sun","year":"2017","journal-title":"Proceedings of the IEEE International Conference on Computer Vision"},{"key":"B26","doi-asserted-by":"publisher","first-page":"2262","DOI":"10.1109\/CVPR.2009.5206705","article-title":"\u201cWhat's it going to cost you? Predicting effort vs. informativeness for multi-label image annotations,\u201d","author":"Vijayanarasimhan","year":"2009","journal-title":"2009 IEEE Conference on Computer Vision and Pattern Recognition"},{"key":"B27","doi-asserted-by":"publisher","first-page":"2591","DOI":"10.1109\/TCSVT.2016.2589879","article-title":"Cost-effective active learning for deep image classification","volume":"27","author":"Wang","year":"2016","journal-title":"IEEE Trans. Circuit Syst. Video Technol"},{"key":"B28","doi-asserted-by":"publisher","first-page":"505","DOI":"10.1109\/ICCVW60793.2023.00057","article-title":"\u201cWeed mapping with convolutional neural networks on high resolution whole-field images,\u201d","author":"Wang","year":"2023","journal-title":"Proceedings of the IEEE\/CVF International Conference on Computer Vision"},{"key":"B29","unstructured":"Yu\n              H.\n            \n            \n              Chen\n              C.\n            \n            \n              Du\n              X.\n            \n            \n              Li\n              Y.\n            \n            \n              Rashwan\n              A.\n            \n            \n              Hou\n              L.\n            \n          \n          Tensorflow model garden\n          \n          2020"}],"container-title":["Frontiers in Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2024.1498956\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,1,9]],"date-time":"2025-01-09T06:13:07Z","timestamp":1736403187000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/frai.2024.1498956\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,1,9]]},"references-count":29,"alternative-id":["10.3389\/frai.2024.1498956"],"URL":"https:\/\/doi.org\/10.3389\/frai.2024.1498956","relation":{},"ISSN":["2624-8212"],"issn-type":[{"value":"2624-8212","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,1,9]]},"article-number":"1498956"}}