CFA: Coupled-Hypersphere-Based Feature Adaptation for Target-Oriented Anomaly Localization

Abstract

For a long time, anomaly localization has been widely used in industries. Previous studies focused on approximating the distribution of normal features without adaptation to a target dataset. However, since anomaly localization should precisely discriminate normal and abnormal features, the absence of adaptation may make the normality of abnormal features overestimated. Thus, we propose Coupled-hypersphere-based Feature Adaptation (CFA) which accomplishes sophisticated anomaly localization using features adapted to the target dataset. CFA consists of (1) a learnable patch descriptor that learns and embeds target-oriented features and (2) scalable memory bank independent of the size of the target dataset. And, CFA adopts transfer learning to increase the normal feature density so that abnormal features can be clearly distinguished by applying patch descriptor and memory bank to a pretrained CNN. The proposed method outperforms the previous methods quantitatively and qualitatively. For example, it provides an AUROC score of 99.5% in anomaly detection and 98.5% in anomaly localization of MVTec AD benchmark. In addition, this paper points out the negative effects of biased features of pre-trained CNNs and emphasizes the importance of the adaptation to the target dataset. The code is publicly available at https://github.com/ sungwool/CFA_for_anomaly_localization

Deep Industrial Image Anomaly Detection: A Survey

“…PatchCore [61] -ResNet The paper introduces a core-set sampling method to build a memory bank. CFA [62] SVDD ResNet The paper improves PatchCore so that image features are distributed on a hypersphere. FAPM [63] -ResNet The paper puts different position features of the image into different memory banks to speed up retrieval.…”

“…Since PatchCore was proposed, numerous improved methods have been developed on its foundation. Coupled-hypersphere-based Feature Adaptation (CFA) is proposed by Lee et al [62] to obtain target-oriented features. The center and surface of the hypersphere in the memory bank are obtained through transfer learning, and the positional relationship between the test feature and the coupled-hypersphere can be used to determine whether it is abnormal or not.…”

DiffusionAD: Denoising Diffusion for Anomaly Detection

“…Anomaly Detection. Modern methods for anomaly detection can be divided into two main paradigms, namely deep feature embedding-based approaches [10,11,14,26,27,37] and generative model-based approaches [19,34,39,59,60].…”

“…These methods in- clude but are not limited to, deep feature embeddings and generative models. Deep feature embedding-based methods [10,11,14,19,26,37,39] often suffer from degraded performance when the distribution of industrial images differs significantly from the one used for feature extraction, as they rely on pre-trained feature extractors on extra datasets such as ImageNet. Generative model-based methods [1,12,34,44,60] require no extra data and are widely applicable in various scenarios.…”

N-pad : Neighboring Pixel-based Industrial Anomaly Detection

“…In addition, attempts have been made to adapt the weights of the pre-trained model to identify the distribution of nominal data. FastFlow [40], FEFM [37], and CFLOW-AD [15] reported good performances by estimating the distribution of network-based features by normalizing the flow, and CFA [19] implemented feature adaption through Coupled-hypersphere to better ex-plain the distribution of nominal features.…”