Ensuring reliable confidence scores from deep neural networks is of paramount significance in critical decision-making systems, particularly in real-world domains such as healthcare. Recent literature on calibrating deep segmentation networks has resulted in substantial progress. Nevertheless, these approaches are strongly inspired by the advancements in classification tasks, and thus their uncertainty is usually modeled by leveraging the information of individual pixels, disregarding the local structure of the object of interest. Indeed, only the recent Spatially Varying Label Smoothing (SVLS) approach considers pixel spatial relationships across classes, by softening the pixel label assignments with a discrete spatial Gaussian kernel. In this work, we first present a constrained optimization perspective of SVLS and demonstrate that it enforces an implicit constraint on soft class proportions of surrounding pixels. Furthermore, our analysis shows that SVLS lacks a mechanism to balance the contribution of the constraint with the primary objective, potentially hindering the optimization process. Based on these observations, we propose NACL (Neighbor Aware CaLibration), a principled and simple solution based on equality constraints on the logit values, which enables to control explicitly both the enforced constraint and the weight of the penalty, offering more flexibility. Comprehensive experiments on a wide variety of well-known segmentation benchmarks demonstrate the superior calibration performance of the proposed approach, without affecting its discriminative power. Furthermore, ablation studies empirically show the model agnostic nature of our approach, which can be used to train a wide span of deep segmentation networks.

翻译：确保深度神经网络输出的置信度分数可靠，在关键决策系统（尤其是医疗等实际应用领域）中具有至关重要的意义。近年来，深度分割网络校准研究取得了显著进展，但这些方法很大程度上受分类任务进展的启发，其不确定性通常通过利用单个像素信息建模，忽略目标对象的局部结构。实际上，仅有最近提出的空间变分标签平滑（SVLS）方法通过使用离散空间高斯核软化像素标签分配，考虑了跨类别的像素空间关系。本文首先从约束优化角度分析SVLS，证明其隐含地对周围像素的软类比例施加约束。进一步分析表明，SVLS缺乏平衡约束项与主目标贡献的机制，可能阻碍优化过程。基于这些发现，我们提出NACL（邻域感知校准）——一种基于logit值等式约束的理论严谨且简洁的解决方案，能够显式控制约束项强度与惩罚权重，提供更高灵活性。在多种经典分割基准上的全面实验表明，该方法在不影响判别能力的前提下实现了优异的校准性能。此外，消融研究从经验角度证实了本方法的模型无关特性，可广泛用于训练各类深度分割网络。