Local Temperature Scaling for Probability Calibration
Local Temperature Scaling for Probability Calibration
复制标题
DOI：
10.1109/iccv48922.2021.00681
复制
发表时间：
2020-08
期刊：
影响因子：
--
通讯作者：
Zhipeng Ding;Xu Han;Peirong Liu;M. Niethammer
中科院分区：
文献类型：
--
作者：
Zhipeng Ding;Xu Han;Peirong Liu;M. Niethammer
对于语义分割，标签概率通常未经校准，因为它们通常只是分割任务的副产品。并交交集 (IoU) 和 Dice 得分通常用作分割成功的标准，而与标签概率相关的指标并不经常被探索。然而，已经研究了概率校准方法，将概率输出与实验观察到的误差相匹配。这些方法主要关注分类任务，而不是语义分割。因此，我们提出了一种基于学习的校准方法，专注于多标签语义分割。具体来说，我们采用卷积神经网络来预测局部温度值以进行概率校准。我们的方法的一个优点是它不会改变预测精度，因此允许将校准作为后处理步骤。 COCO、CamVid 和 LPBA40 数据集上的实验证明了一系列不同指标的校准性能得到了改善。我们还展示了我们的方法从磁共振图像进行多图谱大脑分割的良好性能。
For semantic segmentation, label probabilities are often uncalibrated as they are typically only the by-product of a segmentation task. Intersection over Union (IoU) and Dice score are often used as criteria for segmentation success, while metrics related to label probabilities are not often explored. However, probability calibration approaches have been studied, which match probability outputs with experimentally observed errors. These approaches mainly focus on classification tasks, but not on semantic segmentation. Thus, we propose a learning-based calibration method that focuses on multi-label semantic segmentation. Specifically, we adopt a convolutional neural network to predict local temperature values for probability calibration. One advantage of our approach is that it does not change prediction accuracy, hence allowing for calibration as a postprocessing step. Experiments on the COCO, CamVid, and LPBA40 datasets demonstrate improved calibration performance for a range of different metrics. We also demonstrate the good performance of our method for multi-atlas brain segmentation from magnetic resonance images.