Deep Metric Learning for Open World Semantic Segmentation

• DOI: 10.48550/arxiv.2108.04562
• Publication Date: 2021-01-01
• Authors: Jun Cen; Yun Peng; Junhao Cai; Michael Yu Wang; Ming Liu
• Keywords: FOS: Computer and information sciences; 03 medical and health sciences; 0302 clinical medicine; Computer Vision and Pattern Recognition (cs.CV); Computer Science - Computer Vision and Pattern Recognition; 0202 electrical engineering, electronic engineering, information engineering; 02 engineering and technology

Classical close-set semantic segmentation networks have limited ability to detect out-of-distribution (OOD) objects, which is important for safety-critical applications such as autonomous driving. Incrementally learning these OOD objects with few annotations an ideal way enlarge the knowledge base of deep models. In this paper, we propose open world system that includes two modules: (1) open-set module both in-distribution and objects. (2) incremental few-shot gradually incorporate those into its existing base. This behaves like a human being, able identify learn them corresponding supervision. We adopt Deep Metric Learning Network (DMLNet) contrastive clustering implement segmentation. Compared other methods, our DMLNet achieves state-of-the-art performance on three challenging datasets without using additional data or generative On basis, methods are further proposed progressively improve