Towards Domain Generalization in Object Detection
This addresses the challenge of deploying object detectors in varied real-world environments without needing pre-collected data for all domains, though it is incremental as it builds on existing domain generalization research.
The paper tackles the problem of domain generalization in object detection (DGOD), where detectors trained on source domains must perform on unknown target domains, and proposes a novel method called RAPT that outperforms existing state-of-the-art methods in experiments.
Despite the striking performance achieved by modern detectors when training and test data are sampled from the same or similar distribution, the generalization ability of detectors under unknown distribution shifts remains hardly studied. Recently several works discussed the detectors' adaptation ability to a specific target domain which are not readily applicable in real-world applications since detectors may encounter various environments or situations while pre-collecting all of them before training is inconceivable. In this paper, we study the critical problem, domain generalization in object detection (DGOD), where detectors are trained with source domains and evaluated on unknown target domains. To thoroughly evaluate detectors under unknown distribution shifts, we formulate the DGOD problem and propose a comprehensive evaluation benchmark to fill the vacancy. Moreover, we propose a novel method named Region Aware Proposal reweighTing (RAPT) to eliminate dependence within RoI features. Extensive experiments demonstrate that current DG methods fail to address the DGOD problem and our method outperforms other state-of-the-art counterparts.