发表机构
NAVER Cloud; KAIST（NAVER云; 韩国科学技术院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
提出Phoenix框架，用对抗学习生成语义噪声、对比学习建模细化关系，在图像分割掩码细化任务中性能优于现有方法，可提升先进分割模型。
AI 中文摘要
尽管图像分割领域已取得显著进展，但即使是最先进的模型生成的掩码仍存在边界不完美、语义不一致和结构错误等问题。掩码细化旨在解决这些局限，然而现有方法依赖过于简单的合成噪声，无法捕捉真实分割模型的复杂错误模式。我们提出Phoenix这一新型框架，利用对抗学习生成语义上有意义的噪声模式，并通过对比学习建模细化关系。我们的方法包含两项关键创新：（1）对抗性掩码扰动，采用嵌入攻击创建语义感知噪声，以模拟真实分割错误；（2）对比掩码细化学习，建立三向框架，确保语义区域内的特征一致性，同时保持类别间的区分性。实验表明，Phoenix在各类任务中显著优于现有方法，且能持续提升最先进的分割模型，带来实质性改进。我们的代码和项目页面已公开在该URL。
英文摘要
Despite significant advances in image segmentation, even state-of-the-art models produce masks with imperfect boundaries, semantic inconsistencies, and structural errors. Mask refinement addresses these limitations, yet current approaches rely on simplistic synthetic noise that fails to capture the complex error patterns of real segmentation models. We introduce Phoenix, a novel framework that leverages adversarial learning to generate semantically meaningful noise patterns and contrastive learning to model refinement relationships. Our approach consists of two key innovations: (1) Adversarial Mask Perturbation, which employs embedding attacks to create semantic-aware noise that mimics real segmentation errors, and (2) Contrastive Mask Refinement Learning, which establishes a tri-directional framework that ensures feature consistency within semantic regions while maintaining separation between classes. Experiments demonstrate that Phoenix significantly outperforms existing methods across diverse tasks, while consistently enhancing state-of-the-art segmentation models with substantial improvements. Our code and project page are publicly available at https://phoenix-eccv26.github.io.
CommentsECCV 2026
