Search

Single-image super-resolution (SISR) techniques have found wide applications in semiconductor defect inspection. Enhancing image resolution to improve inspection sensitivity and accuracy holds great significance. A novel SISR algorithm, called cross-convolutional residual network (CCRN), is proposed in this study. CCRN comprises a cross-convolutional module (CCM), which incorporates a cross-sharing mechanism that facilitates the fusion of features from different stages, enabling the extraction of more information from the image. Moreover, a global residual aggregation structure (GRA) is introduced. GRA captures and transfers different levels of residual features acquired from learning each CCM to the reconstruction layer. Experimental results demonstrate that the proposed SR algorithm outperforms existing state-of-the-art SR algorithms in terms of both visual and quantitative metrics when applied to optical, SEM, and TEM images of microfluidic chips, CMOS image sensors, and quantum dots, respectively. Additionally, CCRN significantly improves the accuracy of defect classification and inspection of unpatterned wafers, as evaluated using the WM-811K dataset. Notably, an increase in local defection testing accuracy from 79.00% to 89.00% and an improvement in classification accuracy from 93.69% to 96.06% are achieved. These findings underscore the potential applications of the proposed algorithm in improving semiconductor defect inspection and classification accuracies. [ABSTRACT FROM AUTHOR]

Although expression descriptions provide additional information about facial behaviors despite of different poses, and pose features are beneficial to adapt to pose variety, neither has been fully leveraged in facial expression recognition. This paper proposes a pose-aware text-assisted facial expression recognition method using cross-modality attention. Specifically, the method contains three components. The pose feature extractor extracts pose-related features from facial images, and then cooperates with a fully-connected layer for pose classification. When poses can be clearly discriminated and classified, features obtained from the extractor can represent the corresponding poses. To eliminate bias due to appearance and illumination, cluster centers are taken as the final pose features. The text feature extractor obtains embeddings from expression descriptions. These descriptions are first passed through Intra-Exp attention to obtain preliminary embeddings. To leverage the correlations among expressions, all expression embeddings are then concatenated and passed through Inter-Exp attention. The cross-modality module attempts to learn attention maps that distinguish the importance of facial regions by using prior knowledge about poses and expression descriptions. The image features weighted by the attention maps are utilized to recognize pose and expression jointly. Experiments on three benchmark datasets demonstrate the superiority of the proposed method. [ABSTRACT FROM AUTHOR]