Uncertainty-aware consistency regularization for cross-domain semantic segmentation

被引：0

作者：

Zhou, Qianyu ^{[1
]}

Feng, Zhengyang ^{[1
]}

Gu, Qiqi ^{[1
]}

Cheng, Guangliang ^{[2
]}

Lu, Xuequan ^{[3
]}

Shi, Jianping ^{[2
]}

Ma, Lizhuang ^{[1
]}

机构：

[1] Shanghai Jiao Tong University, 800 Dongchuan Road, Shanghai,200240, China

[2] SenseTime Research, 1900 Hongmei Road, Shanghai,200233, China

[3] Deakin University, 75 Pigdons Rd, Waurn Ponds,VIC,3216, Australia

来源：

Computer Vision and Image Understanding | 2022年 / 221卷

基金：

中国国家自然科学基金;

关键词：

Consistency regularization - Cross-domain - Domain adaptation - Domain semantics - Regularisation - Semantic segmentation - Target domain - Teacher models - Transfer learning - Uncertainty;

D O I：

暂无

中图分类号：

学科分类号：

摘要：

Unsupervised domain adaptation (UDA) aims to adapt existing models of the source domain to a new target domain with only unlabeled data. Most existing methods suffer from noticeable negative transfer resulting from either the error-prone discriminator network or the unreasonable teacher model. Besides, the local regional consistency in UDA has been largely neglected, and only extracting the global-level pattern information is not powerful enough for feature alignment due to the abuse use of contexts. To this end, we propose an uncertainty-aware consistency regularization method for cross-domain semantic segmentation. Firstly, we introduce an uncertainty-guided consistency loss with a dynamic weighting scheme by exploiting the latent uncertainty information of the target samples. As such, more meaningful and reliable knowledge from the teacher model can be transferred to the student model. We further reveal the reason why the current consistency regularization is often unstable in minimizing the domain discrepancy. Besides, we design a ClassDrop mask generation algorithm to produce strong class-wise perturbations. Guided by this mask, we propose a ClassOut strategy to realize effective regional consistency in a fine-grained manner. Experiments demonstrate that our method outperforms the state-of-the-art methods on four domain adaptation benchmarks, i.e., GTAV → Cityscapes, SYNTHIA → Cityscapes, Virtual KITTI ⟶ KITTI and Cityscapes ⟶ KITTI. © 2022 Elsevier Inc.

引用