DA-Cal: Towards Cross-Domain Calibration in Semantic Segmentation

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Li, Wangkai, Sun, Rui, Li, Zhaoyang, Chen, Yujia, Zhang, Tianzhu
Format:	Preprint
Published:	2026
Subjects:	Computer Vision and Pattern Recognition
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

_version_	1866911464994897920
author	Li, Wangkai Sun, Rui Li, Zhaoyang Chen, Yujia Zhang, Tianzhu
author_facet	Li, Wangkai Sun, Rui Li, Zhaoyang Chen, Yujia Zhang, Tianzhu
contents	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. The code will be released.
format	Preprint
id	arxiv_https___arxiv_org_abs_2602_20860
institution	arXiv
publishDate	2026
record_format	arxiv
spellingShingle	DA-Cal: Towards Cross-Domain Calibration in Semantic Segmentation Li, Wangkai Sun, Rui Li, Zhaoyang Chen, Yujia Zhang, Tianzhu Computer Vision and Pattern Recognition 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. The code will be released.
title	DA-Cal: Towards Cross-Domain Calibration in Semantic Segmentation
topic	Computer Vision and Pattern Recognition
url	https://arxiv.org/abs/2602.20860