Do We Need Perfect Data? Leveraging Noise for Domain Generalized Segmentation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kim, Taeyeong, Lee, SeungJoon, Kim, Jung Uk, Cho, MyeongAh
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914173711024128
author Kim, Taeyeong
Lee, SeungJoon
Kim, Jung Uk
Cho, MyeongAh
author_facet Kim, Taeyeong
Lee, SeungJoon
Kim, Jung Uk
Cho, MyeongAh
contents Domain generalization in semantic segmentation faces challenges from domain shifts, particularly under adverse conditions. While diffusion-based data generation methods show promise, they introduce inherent misalignment between generated images and semantic masks. This paper presents FLEX-Seg (FLexible Edge eXploitation for Segmentation), a framework that transforms this limitation into an opportunity for robust learning. FLEX-Seg comprises three key components: (1) Granular Adaptive Prototypes that captures boundary characteristics across multiple scales, (2) Uncertainty Boundary Emphasis that dynamically adjusts learning emphasis based on prediction entropy, and (3) Hardness-Aware Sampling that progressively focuses on challenging examples. By leveraging inherent misalignment rather than enforcing strict alignment, FLEX-Seg learns robust representations while capturing rich stylistic variations. Experiments across five real-world datasets demonstrate consistent improvements over state-of-the-art methods, achieving 2.44% and 2.63% mIoU gains on ACDC and Dark Zurich. Our findings validate that adaptive strategies for handling imperfect synthetic data lead to superior domain generalization. Code is available at https://github.com/VisualScienceLab-KHU/FLEX-Seg.
format Preprint
id arxiv_https___arxiv_org_abs_2511_22948
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Do We Need Perfect Data? Leveraging Noise for Domain Generalized Segmentation
Kim, Taeyeong
Lee, SeungJoon
Kim, Jung Uk
Cho, MyeongAh
Computer Vision and Pattern Recognition
Domain generalization in semantic segmentation faces challenges from domain shifts, particularly under adverse conditions. While diffusion-based data generation methods show promise, they introduce inherent misalignment between generated images and semantic masks. This paper presents FLEX-Seg (FLexible Edge eXploitation for Segmentation), a framework that transforms this limitation into an opportunity for robust learning. FLEX-Seg comprises three key components: (1) Granular Adaptive Prototypes that captures boundary characteristics across multiple scales, (2) Uncertainty Boundary Emphasis that dynamically adjusts learning emphasis based on prediction entropy, and (3) Hardness-Aware Sampling that progressively focuses on challenging examples. By leveraging inherent misalignment rather than enforcing strict alignment, FLEX-Seg learns robust representations while capturing rich stylistic variations. Experiments across five real-world datasets demonstrate consistent improvements over state-of-the-art methods, achieving 2.44% and 2.63% mIoU gains on ACDC and Dark Zurich. Our findings validate that adaptive strategies for handling imperfect synthetic data lead to superior domain generalization. Code is available at https://github.com/VisualScienceLab-KHU/FLEX-Seg.
title Do We Need Perfect Data? Leveraging Noise for Domain Generalized Segmentation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2511.22948