解决跨域分割中置信度不准问题，提升安全关键场景可靠性。
DA-Cal: Towards Cross-Domain Calibration in Semantic Segmentation
- 通过软伪标签优化实现目标域校准，引入元温度网络生成像素级参数。
- 在多个基准上显著改善校准性能，且不增加推理开销。
- 适合对预测可信度要求高的自动驾驶、医疗影像等安全敏感领域。
现有无监督域适应（UDA）方法虽能显著提升语义分割在目标域的表现，但常忽视网络校准质量，导致预测置信度与实际准确率不匹配，这在安全关键应用中构成重大风险。我们观察到，在跨域场景下，使用软伪标签替代硬伪标签会导致性能大幅下降，尽管理论上完美校准的软伪标签应等同于硬伪标签。基于此发现，我们提出 DA-Cal，一个专用于跨域校准的框架，将目标域校准转化为软伪标签优化问题。DA-Cal 引入元温度网络生成像素级校准参数，并采用双层优化建立软伪标签与 UDA 监督信号之间的关系，同时结合互补域混合策略防止过拟合并降低域间差异。实验表明，DA-Cal 可无缝集成至多个现有自训练框架，在多个 UDA 分割基准上显著提升目标域校准效果，且无需额外推理开销。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。