Local Temperature Scaling for Probability Calibration

By Zhipeng Ding, Xu Han, Peirong Liu, Marc Niethammer. 2020

122 citations recorded by the source corpus.

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.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Evaluation & Benchmarks, Probabilistic & Bayesian Methods

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

  1. Marc Niethammer
  2. Terrence Chen
  3. Ziyan Wu
  4. Qin Liu

In Computer Vision and Pattern Recognition

  1. Qin Liu
  2. Marc Niethammer
  3. Terrence Chen
  4. Srikrishna Karanam
  5. Meng Zheng
  6. Ziyan Wu
  7. Benjamin Planche
  8. Jitendra Malik

In Evaluation & Benchmarks

  1. Marc Niethammer
  2. Ziyan Wu
  3. Qin Liu

In Probabilistic & Bayesian Methods

  1. Marc Niethammer

Citations in the corpus