Domain Adaptation for Object Detection using SE Adaptors and Center Loss
This work addresses the practical problem of domain shift in object detection for automotive systems, offering an incremental improvement over existing methods.
The paper tackles the problem of cross-domain robustness in object detection for automotive applications by introducing an unsupervised domain adaptation method that addresses domain shift at instance and image levels with consistency regularization, SE Adaptors for domain attention, and center loss for intra-class variance. The method outperforms previous baselines on Cityscapes to Foggy Cityscapes adaptation.
Despite growing interest in object detection, very few works address the extremely practical problem of cross-domain robustness especially for automative applications. In order to prevent drops in performance due to domain shift, we introduce an unsupervised domain adaptation method built on the foundation of faster-RCNN with two domain adaptation components addressing the shift at the instance and image levels respectively and apply a consistency regularization between them. We also introduce a family of adaptation layers that leverage the squeeze excitation mechanism called SE Adaptors to improve domain attention and thus improves performance without any prior requirement of knowledge of the new target domain. Finally, we incorporate a center loss in the instance and image level representations to improve the intra-class variance. We report all results with Cityscapes as our source domain and Foggy Cityscapes as the target domain outperforming previous baselines.