用偏好学习生成更真实多样的异常图像，无需人工标注。
Anomaly-Preference Image Generation
- 将异常生成转为偏好学习，利用真实异常做正样本指导优化。
- 动态分配模型能力，高噪声阶段保多样性，低噪声阶段提细节还原度。
- 可灵活控制生成一致性与对齐度，适合工业缺陷检测场景。
从有限数据中合成真实且多样化的异常样本对提升模型泛化能力至关重要。然而，现有方法难以兼顾生成质量与多样性，分别受分布错位和过拟合困扰。为此，我们提出异常偏好优化（Anomaly Preference Optimization），将异常生成重构为偏好学习问题。核心是隐式偏好对齐机制，利用真实异常作为正参考，通过去噪轨迹偏差直接获取优化信号，无需昂贵的人工标注。此外，提出时间感知容量分配模块，沿扩散时间轴动态分配模型能力：在高噪声阶段优先保障结构多样性，在低噪声阶段增强细粒度真实性。推理时采用分层采样策略，调节一致性与对齐性的权衡，实现精确控制。大量实验表明，该方法显著优于现有基线，在真实性和多样性上均达到当前最优性能。
原文摘要 · Abstract (English)
Synthesizing realistic and diverse anomalous samples from limited data is vital for robust model generalization. However, existing methods struggle to reconcile fidelity and diversity, often hampered by distribution misalignment and overfitting, respectively.To mitigate this, we introduce Anomaly Preference Optimization,a novel paradigm that reformulates anomaly generation as a preference learning problem.Central to our approach is an implicit preference alignment mechanism that leverages real anomalies as positive references, deriving optimization signals directly from denoising trajectory deviations without requiring costly human annotation. Furthermore, we propose a Time-Aware Capacity Allocation module that dynamically distributes model capacity along the diffusion timeline,prioritizing structural diversity during highnoise phases while enhancing fine-grained fidelity in low-noise stages. During inference, a hierarchical sampling strategy modulates the coherencealignment trade-off, enabling precise control over generation. Extensive experiments demonstrate that significantly outperforms existing baselines,achieving state-of-the-art performance in both realism and diversity.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。