2021/12/27 by Zu-Yun Shiau, Weiwei Lin, Shiau, Zu-Yun +6
Biochemistry, Genetics and Molecular Biology · Computer Science · Mathematics · #Artificial intelligence #Benchmark (surveying) #Cancer-related molecular mechanisms research #Cartography #Classifier (UML) #Computer Vision and Pattern Recognition (cs.CV) #Computer science #Domain (mathematical analysis) #Domain Adaptation and Few-Shot Learning #Domain adaptation #FOS: Computer and information sciences #Feature (linguistics) #Generalization #Geography #Machine learning #Mathematics #Multimodal Machine Learning Applications #Pattern recognition (psychology) #Robustness (evolution) #Segmentation #cs.CV
paper · pdf · doi:10.48550/arxiv.2112.13538
published in arXiv (Cornell University) (Cornell University) · Accepted by ICIP 2021
arxiv created 2021/12/27 · openalex publication_date 2021/12/27 · arxiv updated 2021/12/28 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/08
How to handle domain shifts when recognizing or segmenting visual data across\ndomains has been studied by learning and vision communities. In this paper, we\naddress domain generalized semantic segmentation, in which the segmentation\nmodel is trained on multiple source domains and is expected to generalize to\nunseen data domains. We propose a novel meta-learning scheme with feature\ndisentanglement ability, which derives domain-invariant features for semantic\nsegmentation with domain generalization guarantees. In particular, we introduce\na class-specific feature critic module in our framework, enforcing the\ndisentangled visual features with domain generalization guarantees. Finally,\nour quantitative results on benchmark datasets confirm the effectiveness and\nrobustness of our proposed model, performing favorably against state-of-the-art\ndomain adaptation and generalization methods in segmentation.\n