Abstract
While existing unsupervised domain adaptation (UDA) methods greatly enhance target domain performance in semantic segmentation, they often neglect network calibration quality, resulting in misalignment between prediction confidence and actual accuracy -- a significant risk in safety-critical applications. Our key insight emerges from observing that performance degrades substantially when soft pseudo-labels replace hard pseudo-labels in cross-domain scenarios due to poor calibration, despite the theoretical equivalence of perfectly calibrated soft pseudo-labels to hard pseudo-labels. Based on this finding, we propose DA-Cal, a dedicated cross-domain calibration framework that transforms target domain calibration into soft pseudo-label optimization. DA-Cal introduces a Meta Temperature Network to generate pixel-level calibration parameters and employs bi-level optimization to establish the relationship between soft pseudo-labels and UDA supervision, while utilizing complementary domain-mixing strategies to prevent overfitting and reduce domain discrepancies. Experiments demonstrate that DA-Cal seamlessly integrates with existing self-training frameworks across multiple UDA segmentation benchmarks, significantly improving target domain calibration while delivering performance gains without inference overhead. The code will be released.
Abstract (translated)
虽然现有的无监督领域适应（UDA）方法在语义分割的目标域性能上有了显著的提升，但它们通常忽略了网络校准质量的问题，导致预测置信度与实际准确性之间存在不一致——这在安全关键的应用中是一个重大风险。我们的主要见解来源于观察到，在跨域场景中当软伪标签取代硬伪标签时，尽管理论上完美校准的软伪标签等同于硬伪标签，但由于校准不佳，性能会显著下降。 基于这一发现，我们提出了DA-Cal，这是一种专门针对跨域校准的框架，它将目标领域的校准转化为对软伪标签的优化。DA-Cal 引入了一个元温度网络来生成像素级校准参数，并采用双层优化方法建立软伪标签与UDA监督之间的关系，同时利用互补的领域混合策略来防止过拟合并减少领域差异。 实验表明，DA-Cal 可以无缝集成到现有的自我训练框架中，在多个UDA分割基准测试上显著提升了目标域的校准质量，而不会增加推理开销，并且提高了性能。代码将对外发布。
URL
https://arxiv.org/abs/2602.20860
