CD-FKD: Cross-Domain Feature Knowledge Distillation for Robust Single-Domain Generalization in Object Detection

TL;DR

This paper introduces CD-FKD, a novel cross-domain feature knowledge distillation method that significantly improves single-domain generalization in object detection by leveraging global and instance-wise feature transfer, especially under challenging domain shifts.

Contribution

It proposes a new distillation framework that enhances model robustness to domain shifts by using diversified data and feature mimicry, outperforming existing methods in challenging scenarios.

Findings

Outperforms state-of-the-art methods in target domain generalization

Improves robustness of object detection under weather and scene variations

Enhances detection of difficult objects in corrupted environments

Abstract

Single-domain generalization is essential for object detection, particularly when training models on a single source domain and evaluating them on unseen target domains. Domain shifts, such as changes in weather, lighting, or scene conditions, pose significant challenges to the generalization ability of existing models. To address this, we propose Cross-Domain Feature Knowledge Distillation (CD-FKD), which enhances the generalization capability of the student network by leveraging both global and instance-wise feature distillation. The proposed method uses diversified data through downscaling and corruption to train the student network, whereas the teacher network receives the original source domain data. The student network mimics the features of the teacher through both global and instance-wise distillation, enabling it to extract object-centric features effectively, even for objects that are difficult to detect owing to corruption. Extensive experiments on challenging scenes demonstrate that CD-FKD outperforms state-of-the-art methods in both target domain generalization and source domain performance, validating its effectiveness in improving object detection robustness to domain shifts. This approach is valuable in real-world applications, like autonomous driving and surveillance, where robust object detection in diverse environments is crucial.