Enregistré dans:
Détails bibliographiques
Auteurs principaux: Guo, Yifu, Lu, Yuquan, Zhang, Wentao, Xu, Zishan, Chen, Dexia, Zhang, Siyu, Zhang, Yizhe, Wang, Ruixuan
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2508.05065
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866917203641630720
author Guo, Yifu
Lu, Yuquan
Zhang, Wentao
Xu, Zishan
Chen, Dexia
Zhang, Siyu
Zhang, Yizhe
Wang, Ruixuan
author_facet Guo, Yifu
Lu, Yuquan
Zhang, Wentao
Xu, Zishan
Chen, Dexia
Zhang, Siyu
Zhang, Yizhe
Wang, Ruixuan
contents Continual Semantic Segmentation (CSS) requires learning new classes without forgetting previously acquired knowledge, addressing the fundamental challenge of catastrophic forgetting in dense prediction tasks. However, existing CSS methods typically employ single-stage encoder-decoder architectures where segmentation masks and class labels are tightly coupled, leading to interference between old and new class learning and suboptimal retention-plasticity balance. We introduce DecoupleCSS, a novel two-stage framework for CSS. By decoupling class-aware detection from class-agnostic segmentation, DecoupleCSS enables more effective continual learning, preserving past knowledge while learning new classes. The first stage leverages pre-trained text and image encoders, adapted using LoRA, to encode class-specific information and generate location-aware prompts. In the second stage, the Segment Anything Model (SAM) is employed to produce precise segmentation masks, ensuring that segmentation knowledge is shared across both new and previous classes. This approach improves the balance between retention and adaptability in CSS, achieving state-of-the-art performance across a variety of challenging tasks. Our code is publicly available at: https://github.com/euyis1019/Decoupling-Continual-Semantic-Segmentation.
format Preprint
id arxiv_https___arxiv_org_abs_2508_05065
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Decoupling Continual Semantic Segmentation
Guo, Yifu
Lu, Yuquan
Zhang, Wentao
Xu, Zishan
Chen, Dexia
Zhang, Siyu
Zhang, Yizhe
Wang, Ruixuan
Computer Vision and Pattern Recognition
Continual Semantic Segmentation (CSS) requires learning new classes without forgetting previously acquired knowledge, addressing the fundamental challenge of catastrophic forgetting in dense prediction tasks. However, existing CSS methods typically employ single-stage encoder-decoder architectures where segmentation masks and class labels are tightly coupled, leading to interference between old and new class learning and suboptimal retention-plasticity balance. We introduce DecoupleCSS, a novel two-stage framework for CSS. By decoupling class-aware detection from class-agnostic segmentation, DecoupleCSS enables more effective continual learning, preserving past knowledge while learning new classes. The first stage leverages pre-trained text and image encoders, adapted using LoRA, to encode class-specific information and generate location-aware prompts. In the second stage, the Segment Anything Model (SAM) is employed to produce precise segmentation masks, ensuring that segmentation knowledge is shared across both new and previous classes. This approach improves the balance between retention and adaptability in CSS, achieving state-of-the-art performance across a variety of challenging tasks. Our code is publicly available at: https://github.com/euyis1019/Decoupling-Continual-Semantic-Segmentation.
title Decoupling Continual Semantic Segmentation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2508.05065