Automated inspection using deep-learning has been attracting attention for visual inspection at the manufacturing site. However, the inability to obtain sufficient abnormal product data for training deep- learning models is a problem in practical application. This study proposes an anomaly detection method based on the Siamese network with an attention mechanism for a small dataset. Moreover, attention branch loss (ABL) is proposed for Siamese network to render more task-specific attention maps from attention mechanism. Experimental results confirm that the proposed method with the attention mechanism and ABL is effective even with limited abnormal data.
3 November 2020 in Applied Artificial Intelligence
In this paper, we explore how higher-level perceptual information based on visual attention can be used for aesthetic assessment of images. We assume that visually dominant subjects in a photograph influence stronger aesthetic interest. In other words, visual attention may be a key to predicting photographic aesthetics. Our proposed aesthetic assessment method, which is based on multi-stream and multi-task convolutional neural networks (CNNs), extracts global features and saliency features from an input image. These provide higher-level visual information such as the quality of the photo subject and the subject–background relationship. Results from our experiments support the effectiveness of our approach.
Unbalanced nutrition due to an unhealthy diet may increase the risk of developing lifestyle diseases. Many mobile applications have been released to record everyday meals for the health-conscious to enable them to improve their dietary habits. Most of these applications only base their food classification on an image of the food, requiring the user to manually input information about the ingredients such as the calories and salinity. To address this problem, food ingredient estimation from food images has been attracting increasing attention. Automatic ingredient estimation could possibly strongly alleviate the process of food-intake estimation and dietary assessment. In this paper, we propose an automatic food ingredient estimation method from food images by using multi-task CNN. We focus on classification of the food category and estimation of the calorie content and salinity for lifestyle disease prevention. Two-stage transfer learning using a large number of food category recognition image databases is applied to train our multi-task CNN for improved ingredient estimation. We experimentally analyze the relationship between the food category and salinity by using multi-task CNN.
The purpose of image aesthetics assessment is to automatically predict the perceived quality of an image. Convolutional neural network (CNN) based on deep learning has been used for aesthetics assessment and has displayed potential results. Our final objective is to identify features that contribute to the estimation of aesthetic quality by using explainable AI. In this study, we focused on composition as the first step. By applying clustering to the attention maps obtained by CNN and Grad-CAM++, it was experimentally verified whether the CNN model considers composition for aesthetics assessment. In addition, we verified whether the aesthetic quality features that humans pay attention to differ according to photographic categories such as landscape or portrait.