arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2303.09083cs.CV
收藏

专注于目标:领域自适应语义分割的师生双重框架

Focus on Your Target: A Dual Teacher-Student Framework for Domain-adaptive Semantic Segmentation

University of Science and Technology of China(中国科学技术大学);Huawei Inc.(华为公司)

Xinyue Huo, Lingxi Xie, Wengang Zhou, Houqiang Li, Qi Tian

摘要:

我们研究了无监督域自适应(UDA)的语义分割。目前,流行的UDA框架在于自训练,其赋予模型双重能力:(i)从源域中的标记图像学习可靠的语义,以及(ii)通过在未标记图像上生成伪标记来适应目标域。我们发现,通过减少/增加来自目标域的训练样本的比例,“学习能力”被加强/减弱,而“适应能力”则相反,这意味着这两种能力之间的冲突,特别是对于单个模型。为了缓解这个问题,我们提出了一个新的双重教师-学生(ESTA)的框架,并配备了双向学习策略。通过增加目标领域数据的比例,第二个师生模型学会了“专注于你的目标”,而第一个模型不受影响。它可以很容易地插入到现有的自我训练方法中。在标准UDA场景中(在合成的标记数据和真实的未标记数据上进行训练),GTAv在基线上显示出一致的增益,并在GTAv $\rightarrow$ Cityscapes和SYNTHIA $\rightarrow$ Cityscapes上分别设置了76.5\ %和75.1\ % mIoU的最新结果。

英文摘要:

We study unsupervised domain adaptation (UDA) for semantic segmentation. Currently, a popular UDA framework lies in self-training which endows the model with two-fold abilities: (i) learning reliable semantics from the labeled images in the source domain, and (ii) adapting to the target domain via generating pseudo labels on the unlabeled images. We find that, by decreasing/increasing the proportion of training samples from the target domain, the 'learning ability' is strengthened/weakened while the 'adapting ability' goes in the opposite direction, implying a conflict between these two abilities, especially for a single model. To alleviate the issue, we propose a novel dual teacher-student (DTS) framework and equip it with a bidirectional learning strategy. By increasing the proportion of target-domain data, the second teacher-student model learns to 'Focus on Your Target' while the first model is not affected. DTS is easily plugged into existing self-training approaches. In a standard UDA scenario (training on synthetic, labeled data and real, unlabeled data), DTS shows consistent gains over the baselines and sets new state-of-the-art results of 76.5\% and 75.1\% mIoUs on GTAv$\rightarrow$Cityscapes and SYNTHIA$\rightarrow$Cityscapes, respectively.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑