In construction sites, construction machinery such as excavators plays a critical role. The management of such equipment, notably the monitoring of actions conducted by each construction machinery, is, therefore, key to high productivity and efficiency. This time-consuming and laborious task is currently conducted manually by humans and thus, its automation is highly sought after. Previous works on this issue have achieved high performance using deep learning-based approaches and cameras. However, the investments needed to obtain the training data critical to such approaches are often prohibitive. Using a simulator to generate the training data appears therefore as an alternative to allow fast and easy gathering of training data. However, models trained using such training data perform poorly on real data. The purpose of this study is therefore to increase the performance of action recognition of construction machinery such as excavators using simulator-generated training data. A data augmentation process using background images gathered from actual construction sites is used to reduce the gap between simulator-generated data and real-world data. Experiments with data collected in an actual construction site showed the effectiveness of the proposed method.

Facial expressions are expressed by subtle changes in the shape of the eyes and mouth, wrinkles between the eyebrows, etc. However, it is difficult to capture these changes. In this study, we propose a Recurrent Attention Module (RAM) and a facial expression recognition method using RAM to capture subtle changes in facial expressions. In our experiments, we used CK+ and eNTEFRACE05 databases. In CK+, the recognition accuracies of ConvLSTM and Enhanced ConvLSTM are 93.0% and 95.7%, respectively, while the addition of RAM improves the accuracies by 2.7% and 1.2%, respectively. Furthermore, the recognition accuracies of ConvLSTM and Enhanced ConvLSTM in the eNTERFACE05 database were 39.8% and 49.3%, respectively, while adding RAM improved the accuracies by 4.6% and 0.6%, respectively. In comparison with the conventional method, the proposed method could not outperform the conventional method in CK+. On the other hand, the proposed method improves the accuracy of the eNTERFACE05 database by 0.6% over the conventional method.

A mechanism was designed to synergize theoretical and computer-aided engineering (CAE) education using Simscape Multibody developed by Mathworks Inc. Educational outcomes were examined by observing students' scores in theory and CAE assignments after each class via decision tree analysis and Bayesian networks. The decision tree analysis showed that the scores on a calculation of gear dimensions influenced the grades of the students the most and were related to their understanding of CAE. Students with high assignment scores were eventually able to model and simulate a mechanism described in a technical book, which was a synergistic effect of theoretical and CAE education. However, it was observed that students with poor assignment scores were inept in the CAE learning process, leading up to the quadric crank chain mechanism. The Bayesian networks showed low dependency between classes in the CAE learning process, which implied that the intended educational outcomes were not achieved. Based on these analyses, which were demonstrated to be effective in evaluating class design, the curriculum, time allocation, and follow-up system of the educational process will be improved and classes will be conducted effectively.

This paper presents the method by which binarize Printed Wiring Board(PWB) images. Multilayer printed wired boards have many wiring layers inside the board, and the images capturing these wiring layers contains a large number of scratches and stains, because the wiring layer images are captured while the surface is thinly scraped. We aim to recover the wiring pattern as a binarized image from such images. We have devised an input layer based on the U-Net structure, which consists of 1) 3-channel RGB input (conventional U-Net), 2) 5-channel input by adding L and S channels of HLS color space to RGB, 3) 4-channel input by adding a binary image by Otsu's method of L to RGB, 4) a loop that repeats the structure of method 3), and compared them. As a result, we found that method 3) with Otsu's binarization added to the input was superior in all the indices of accuracy, F-measure, mIoU, and the number of breaks and short circuits.

Wafer defect inspection is essential to control quality and yield of semiconductor device manufacturing. Wafer surface inspection tools with laser scattering method are employed for high-throughput and high-sensitivity inspection of defects with sizes of about several microns to sub-20 nm. In addition to detecting defects with various shapes and sizes, classification of defect types is required. Classification by multi-directional scattered light detection is known, which captures defect scattered light at multiple scatter direction and compares the intensities in different angles. We simulate concave-convex defect classification with multi-directional detection by analyzing polarized angular scattering distributions, obtained with discrete dipole approximation method, from defects that differ in shapes, sizes, and materials. As a result of classification performance evaluation with support vector machine classifier, when using only unpolarized light intensities, the maximum defect diameter that is accurately classified is limited to from 0.62 times as much as the illumination wavelength. By using both +45 and -45 deg linearly polarized light intensities, the limit is extended to from 1.47 times as much as the wavelength.

With the recent increase in the size and image quality of optical equipment, there is a strong demand for higher precision aspherical lenses. Generally, an aspherical glass lens or its mold is formed by grinding and then finished by polishing. Since the shape accuracy deteriorates in this polishing process, it is necessary to keep the polishing amount small. Therefore, it is desirable to optimize the grinding conditions and reduce the grinding surface roughness. In this paper, to optimize the grinding conditions of axisymmetric aspherical surface, maximum height roughness of axisymmetric aspherical ground surface is analyzed theoretically utilizing the statistical grinding theory. From the view point of relationship between the grain cutting direction and workpiece feed direction, grinding can be classified into parallel grinding and cross grinding. And it is found that the parallel grinding is suited to axisymmetric aspherical grinding.

This paper deals with the measurement and simulation of the saw wire temperature during machining a rock sample in vacuum. It is expected for wire-sawing to keep cutting performance by successive supply of cutting edges. The performance of wire-sawing in vacuum has been investigated for lunar and planetary explorations. To clarify the deterioration due to the nickel adhesion on the rock, the temperature rise was investigated. In the experiment, the saw wire was reciprocated with slider-clank mechanism to machine a rock sample. A workpiece holder was mounted on a lever though an octagonal load cell with strain gauges to measure the principal force and thrust simultaneously. The saw wire temperature was measured by its electric resistance during machining. In vacuum, the saw wire temperature rose about 30°C at a wire speed of 0.16 m/s. The generated heat during machining was conducted to the rock sample as well as the saw wire. The temperature rise of a saw wire during machining in vacuum was analyzed by a three-dimensional finite difference model. In the simulation, the saw wire temperature rose about 55°C at the same wire speed as the experiment. The trend of the simulated temperature rise coincided with the measured one. The generated heat during machining affected little on the nickel adhesion and the performance of wire-sawing in vacuum.

Aspheric micro-lens arrays have been attracting great attention to improve light distribution control. This study proposes a simple and efficient method for fabricating close-packed aspheric micro-lens array. First, silica particles of ϕ10 μm were self-assembled over ϕ10 mm area using sedimentation method. Then the convex profile of the self-assembled particles was replicated to a spherical micro-lens array mold. The mold material was silicone elastomer or polydimethylsiloxane (PDMS). Then, the mold profile was replicated into ultraviolet (UV) curing resin. By applying uniaxial stretching to the mold during the secondary replication, aspheric micro-lens array was fabricated. Deformation of the PDMS mold under uniaxial stretching was analyzed by using finite element analysis (FEA) for rough estimation. Experimental results showed that intended lens profile was obtained though some defects were observed due to the nonuniformity of the particle diameters. Optical characteristic of elliptic micro-lens array was made clear.

Active learning refers to label-efficient algorithms that use the most representative samples for labeling when creating training data. In this paper, we propose a model that derives the most informative unlabeled samples from the output of a task model. The tasks are a classification problem, multi-label classification and a semantic segmentation problem. The model consists of an uncertainty indicator generator and a task model. After training the task model with labeled samples, the model predicts unlabeled samples, and based on the prediction results, the uncertainty indicator generator outputs an uncertainty indicator for each unlabeled sample. Samples with a higher uncertainty indicator are considered to be more informative and selected. As a result of experiments using multiple datasets, our model achieved better accuracy than conventional active learning methods and reduced execution time by a factor of 10.