发现正常图像中隐藏的近缺陷区域，提升异常检测精度
What Remains Normal? Clean Images Miss Useful Near-Defect Normal Patches for Anomaly Detection
- 用全局覆盖选择法保留更多罕见正常样本
- 新方法使内存污染接近零，12组对比均提升性能
- 适合关注异常检测鲁棒性的研究者使用
基于记忆的异常检测器通过存储正常训练块并在测试时评分。当前方法在固定表示和内存预算下，采用随机、中位数、局部或全局覆盖选择器构建记忆库。我们引入CLEANCON——一种跨图像的袋外支持门机制，在不改变表示、内存大小、构建器和推理规则的前提下，动态调整候选图像资格。实验表明，全局覆盖会过度包含稀疏异常样本，而CLEANCON将最终内存污染降至约零，并在全部12组匹配比较中提升类别宏平均精度-平均精度（P-AP）。然而，在保留率扫描中，最低污染内存并未带来最高P-AP，性能随污染上升持续改善。说明内存污染无法准确反映模型性能排序。代码公开于https://github.com/jw-chae/cleancon。
原文摘要 · Abstract (English)
Memory-based anomaly detectors store nominal training patches and score test patches against this memory. A patch selected for coverage therefore becomes a nor- mal reference without a separate check that geometric rarity makes it safe to trust. We probe this coupling with sparse training contamination. Under fixed representa- tions and memory budgets, we compare random, medoid, local, and global coverage selectors. We then use CLEANCON, an out-of-bag cross-image support gate that changes candidate-image eligibility while fixing the representation, absolute mem- ory size, builder, and inference rule. Global coverage strongly over-represents sparse contamination. CLEANCON reduces final-memory contamination to approx- imately zero and increases category-macro P-AP in all 12 matched comparisons. Yet along a retention sweep, the lowest-contamination memory does not attain the highest P-AP; performance continues to improve while contamination rises. Mem- ory contamination therefore does not order the resulting memories by P-AP.Code is publicly available at https://github.com/jw-chae/cleancon.
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。