Domain Adaptation for Object Detection using SE Adaptors and Center Loss

TL;DR

This paper presents an unsupervised domain adaptation method for object detection that uses SE Adaptors and center loss to improve robustness across domains, demonstrated on Cityscapes and Foggy Cityscapes datasets.

Contribution

It introduces SE Adaptors for better domain attention and incorporates center loss to enhance intra-class variance, advancing cross-domain object detection.

Findings

Outperforms previous baselines on Cityscapes to Foggy Cityscapes adaptation

Utilizes SE Adaptors to improve domain attention without prior target domain knowledge

Incorporates center loss to reduce intra-class variance in representations

Abstract

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.