Abstract:The prevalence of domain adaptive semantic segmentation has prompted concerns regarding source domain data leakage, where private information from the source domain could inadvertently be exposed in the target domain. To circumvent the requirement for source data, source-free domain adaptation has emerged as a viable solution that leverages self-training methods to pseudo-label high-confidence regions and adapt the model to the target data. However, the confidence scores obtained are often highly biased due to over-confidence and class-imbalance issues, which render both model selection and optimization problematic. In this paper, we propose a novel calibration-guided source-free domain adaptive semantic segmentation (Cal-SFDA) framework. The core idea is to estimate the expected calibration error (ECE) from the segmentation predictions, serving as a strong indicator of the model's generalization capability to the unlabeled target domain. The estimated ECE scores, in turn, assist the model training and fair selection in both source training and target adaptation stages. During model pre-training on the source domain, we ensure the differentiability of the ECE objective by leveraging the LogSumExp trick and using ECE scores to select the best source checkpoints for adaptation. To enable ECE estimation on the target domain without requiring labels, we train a value net for ECE estimation and apply statistic warm-up on its BatchNorm layers for stability. The estimated ECE scores assist in determining the reliability of prediction and enable class-balanced pseudo-labeling by positively guiding the adaptation progress and inhibiting potential error accumulation. Extensive experiments on two widely-used synthetic-to-real transfer tasks show that the proposed approach surpasses previous state-of-the-art by up to 5.25% of mIoU with fair model selection criteria.

On Calibrating Semantic Segmentation Models: Analyses and An Algorithm

Troubleshooting image segmentation models with human-in-the-loop

On the Calibration of Uncertainty Estimation in LiDAR-based Semantic Segmentation

Cal-SFDA: Source-Free Domain-adaptive Semantic Segmentation with Differentiable Expected Calibration Error

CALSeg: Improving Calibration of Medical Image Segmentation Via Variational Label Smoothing

Mask-TS Net: Mask Temperature Scaling Uncertainty Calibration for Polyp Segmentation

SFC: Shared Feature Calibration in Weakly Supervised Semantic Segmentation

Diagnostics in Semantic Segmentation

On the Importance of Calibration in Semi-supervised Learning

On Calibration of Pre-trained Code Models

Harmonizing Output Imbalance for semantic segmentation on extremely-imbalanced input data

Renovate Yourself: Calibrating Feature Representation of Misclassified Pixels for Semantic Segmentation.

Open-Vocabulary Segmentation with Semantic-Assisted Calibration

Neighbor-Aware Calibration of Segmentation Networks with Penalty-Based Constraints

Kandinsky Conformal Prediction: Efficient Calibration of Image Segmentation Algorithms

Self-supervised Scale Equivariant Network for Weakly Supervised Semantic Segmentation

Calib3D: Calibrating Model Preferences for Reliable 3D Scene Understanding

Distribution-Aware Margin Calibration for Semantic Segmentation in Images

Confidence Calibration for Object Detection and Segmentation

Calibration in Deep Learning: A Survey of the State-of-the-Art

Semisupervised Semantic Segmentation With Certainty-Aware Consistency Training for Remote Sensing Imagery