Flownomaly improves few-shot anomaly detection with single-step recovery
Flownomaly: Single-Step Conditional-Mean Recovery for Few-Shot Anomaly Detection
Computer Vision and Pattern RecognitionMachine Learning
Summary
Detecting defects in industrial images is hard when only a few normal examples are available. The authors propose Flownomaly, which learns how normal images look in a feature space using a technique that moves a noisy version of an image back to a clean version in one step. This approach helps spot anomalies by comparing the original and cleaned versions without needing examples of defects to tune. Flownomaly achieves high accuracy on several standard benchmarks, improving detection with only one or a few normal images.
What this means in practice
- •For industrial quality control teams: Identify manufacturing defects from only a few defect-free images, improving early anomaly detection without needing faulty samples.
- •For visual inspection system developers: Integrate single-step feature recovery to enhance anomaly localization and image-level detection in production line inspection tools.
Authors
Bernardo Perrone Ribeiro, Erkut Akdag, Egor Bondarev
Abstract
Few-shot industrial anomaly detection requires modeling category-specific normality from only a handful of defect-free images. Common approaches tie inference to stored exemplars, impose rigid linear structure, or risk reproducing anomalous content. Flownomaly instead casts the problem as normal-counterpart recovery in frozen DINOv2 feature space. Given normal supports, a velocity field is learned through conditional flow matching over frozen patch tokens to transport Gaussian noise toward normal features. Along straight paths, a single recovery step using one evaluation of the optimal field yields the posterior mean of the clean normal token without trajectory integration. The operating time governs how strongly recovery modifies a query token. With no anomalous validation samples available for tuning, a normal-only Jacobian gain-to-noise criterion selects this time from the supports alone. Squared residuals between query tokens and their recovered counterparts yield anomaly signals for both dense localization and image-level detection. Across 1-, 2-, and 4-shot benchmarks on MVTec-AD and VisA, together with sample-shot evaluations on Real-IAD, Flownomaly achieves state-of-the-art performance against recent category-specific methods. In the 1-shot setting, image- and pixel-level AUROC reach 97.8% and 98.0% on MVTec-AD and 95.3% and 98.7% on VisA, with a 2.1-point image-level AUROC improvement over the strongest baseline on VisA. Code is available at https://github.com/b-rbmp/Flownomaly.