ProCon: Projection-Consistency Memory for Training-Free Anomaly Detection
2026-07-06 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors propose a new method called ProCon for detecting anomalies in images by comparing patches to a memory bank of normal patches. Unlike previous methods that only check for the nearest match, ProCon softly reconstructs each test patch from nearby normal ones and measures the difference as evidence of anomalies. This approach stabilizes detection without needing extra training or generating fake examples. They tested ProCon on several benchmarks and found it achieved high accuracy in spotting defects.
anomaly detectionmemory banknearest neighborimage reconstructiondecoder-freeprojection residualAUROCMVTec-ADpixel-level detectionsingle-category evaluation
Authors
Joongwon Chae, Lihui Luo, Yang Liu, Dongmei Yu, Peiwu Qin, Runming Wang, Ilmoon Chae
Abstract
Memory-based anomaly detection is attractive because it localizes defects from normal images without training a decoder or synthesizing pseudo anomalies. However, most memory methods still use the memory bank as a nearest-neighbor lookup table: a test patch is treated as normal if it has one nearby normal anchor. This hard retrieval view is vulnerable to false-normal matches and does not test whether the patch is consistently supported by a local normal neighborhood. We propose ProCon, a training-free framework that turns memory retrieval into decoder-free reconstruction. ProCon softly projects each test patch onto nearby normal memory vectors and uses the projection residual as anomaly evidence. To stabilize this residual, it constructs seed-perturbed layer-wise memories, aggregates bank residuals by a median, and fuses depth-specific residual maps by layer consensus. ProCon requires no decoder training, backbone fine-tuning, learned fusion weights, or pseudo-anomaly supervision. Across MVTec-AD, VisA, and Real-IAD under the single-category evaluation protocol, ProCon achieves strong image- and pixel-level performance under seven standard metrics, including image AUROC scores of 99.8%, 99.2%, and 93.2%, respectively. Ablations show that the gains come from replacing hard retrieval with soft normal projection and stabilizing the residuals through memory and depth consensus. The code is available at https://github.com/jw-chae/Procon