arxivcs.CV2026-07-08
Prototype-Anchored Generalized Manifold Regression for Unknown-Domain Object Detection
Zihao Zhang, Aming Wu, Yang Li, Yahong Han
In this paper, we study Single-Domain Generalized Object Detection (Single-DGOD), which aims to transfer a detector trained on a single source domain to multiple unseen domains. Existing methods mainly rely on simulation-driven strategies, such as data augmentation or textual pro…