In this paper, we focus on a realistic yet challenging task, Single Domain Generalization Object Detection (S-DGOD), where only one source domain's data can be used for training object detectors, but have to generalize multiple distinct target domains. In S-DGOD, both high-capacity fitting and generalization abilities are needed due to the task's complexity. Differentiable Neural Architecture Search (NAS) is known for its high capacity for complex data fitting and we propose to leverage Differentiable NAS to solve S-DGOD. However, it may confront severe over-fitting issues due to the feature imbalance phenomenon, where parameters optimized by gradient descent are biased to learn from the easy-to-learn features, which are usually non-causal and spuriously correlated to ground truth labels, such as the features of background in object detection data. Consequently, this leads to serious performance degradation, especially in generalizing to unseen target domains with huge domain gaps between the source domain and target domains. To address this issue, we propose the Generalizable loss (G-loss), which is an OoD-aware objective, preventing NAS from over-fitting by using gradient descent to optimize parameters not only on a subset of easy-to-learn features but also the remaining predictive features for generalization, and the overall framework is named G-NAS. Experimental results on the S-DGOD urban-scene datasets demonstrate that the proposed G-NAS achieves SOTA performance compared to baseline methods. Codes are available at https://github.com/wufan-cse/G-NAS.

翻译：本文聚焦于一个现实且具有挑战性的任务——单域泛化目标检测（Single Domain Generalization Object Detection, S-DGOD），其中仅可使用单个源域的数据训练目标检测器，但需泛化到多个截然不同的目标域。在S-DGOD中，由于任务的复杂性，既需要高容量拟合能力，又需要泛化能力。可微分神经架构搜索（Differentiable Neural Architecture Search, NAS）因其对复杂数据拟合的高容量而著称，我们提出利用可微分NAS解决S-DGOD。然而，由于特征不平衡现象，梯度下降优化的参数倾向于从易学习特征中学习，这些特征通常是非因果的，且与真实标签存在虚假相关性（例如目标检测数据中的背景特征），因此可能面临严重的过拟合问题。这会导致性能严重下降，尤其是在泛化到与源域存在巨大域差距的未见目标域时。为解决这一问题，我们提出可泛化损失（Generalizable loss, G-loss），这是一种面向分布外（OoD）感知的目标函数，通过梯度下降优化参数时不仅关注易学习特征的子集，还保留其预测性特征以实现泛化，从而防止NAS过拟合，整体框架命名为G-NAS。在S-DGOD城市场景数据集上的实验结果表明，与基线方法相比，本文提出的G-NAS达到了最先进的性能。代码已在https://github.com/wufan-cse/G-NAS 开源。