Anomaly detection uses various machine learning techniques to identify and classify defective data on the production line. The autoencoder-based anomaly detection method is an unsupervised method that classifies abnormal samples using an autoencoder trained only from normal samples and is useful in environments where it is difficult to obtain abnormal samples. This method uses an abnormal score based on the reconstruction loss function, making it difficult to detect defects, such as stains, having a similar texture to a normal sample. To solve this problem, we propose an anomaly detection method using a vector quantized variational autoencoder and a feature vector frequency map. We use the prototype vector histogram and its frequency for anomaly detection instead of the reconstruction loss function. The prototype vector histogram is obtained from the vector quantized variational autoencoder's codebook in the training stage. The feature vector frequency map of the input image is generated using the prototype vector histogram in the inference stage. We calculated the abnormal score using the generated frequency map and classified the abnormal samples. The experimental results showed that the proposed method has a higher Area Under Receiver Operating Characteristics (AUROC) than the previous method in stain and scratch defects.

A stain defect is difficult to detect with the human eye because of its characteristic of having a very minimal difference in brightness with the local area of the surface. Recently, with the development of Deep learning, the Convolutional Neural Network (CNN) based stain defect classification method has been proposed. This paper proposes a Dual-stream CNN for stain defect classification using a Gabor filter image. Using Dual-stream structure CNN, Gabor filter images and Gray image (Original) preserve their respective features. The experiment based on the Magnetic Tile (MT) stain data set and the Compact Camera Module (CCM) stain dataset confirms that the proposed method has an improved performance based on the precision, recall, and F1-score in comparison to the Single-stream extraction-based method. Gabor filter images have an advantage in image texture analysis and can be used as an input to the CNNs. The Dual-stream structure better extracts the features needed for classification.