Wu, Tsung HanTsung HanWuLiou, Yi SyuanYi SyuanLiouYuan, Shao JiShao JiYuanLee, Hsin YingHsin YingLeeChen, Tung I.Tung I.ChenKUAN-CHIH HUANGWINSTON HSU2023-07-172023-07-172022-01-01978303119817503029743https://scholars.lib.ntu.edu.tw/handle/123456789/633763https://www.scopus.com/record/display.uri?eid=2-s2.0-85142756162&origin=resultslist&sort=plf-fIn the field of domain adaptation, a trade-off exists between the model performance and the number of target domain annotations. Active learning, maximizing model performance with few informative labeled data, comes in handy for such a scenario. In this work, we present D2ADA, a general active domain adaptation framework for semantic segmentation. To adapt the model to the target domain with minimum queried labels, we propose acquiring labels of the samples with high probability density in the target domain yet with low probability density in the source domain, complementary to the existing source domain labeled data. To further facilitate labeling efficiency, we design a dynamic scheduling policy to adjust the labeling budgets between domain exploration and model uncertainty over time. Extensive experiments show that our method outperforms existing active learning and domain adaptation baselines on two benchmarks, GTA5 → Cityscapes and SYNTHIA → Cityscapes. With less than 5% target domain annotations, our method reaches comparable results with that of full supervision.Active learning | Domain adaptation | Semantic segmentation[SDGs]SDG11conference paper10.1007/978-3-031-19818-2_262-s2.0-85142756162https://api.elsevier.com/content/abstract/scopus_id/85142756162