This paper presents a valid algorithm for a fused sensor, which consists of a passive infrared (PIR) sensor and thermal-image-based sensor, to achieve both energy efficient and privacy in occupancy lighting control. The algorithm efficiently switches between the PIR and thermal-image-based sensors, depending on the lighting state. Additionally, the effectiveness of the fused sensor was verified through an experiment conducted in a single office section for 52.78 hours over eight days. The fused sensor significantly increased the occupancy control accuracy by at least 2.34% and energy-saving effect based on lighting time by at least 1.35% compared to either of its constituent sensors individually. Furthermore, the fused sensor had the lowest comprehensive energy consumption, which was computed as the sum of the energy consumption of both the sensor and light, with a reduction of at least 6.43%. Further comparisons and through discussions are presented in this paper.

Perceptual loss, calculated by VGG network pre-trained on ImageNet, has been widely employed in the past for super-resolution tasks, enabling the generation of photo-realistic images. However, it has been reported that grid-like artifacts frequently appear in the generated images. To address this problem, we consider that large-scale pre-trained models can make significant contributions to super-resolution across different scenes. In particular, by combining language, those models can exhibit a strong capability to comprehend complex scenes, potentially enhancing super-resolution performance. Therefore, this paper proposes new perceptual loss with Contrastive Language-Image Pre-training (CLIP) based on Vision Transformer (ViT) instead of VGG network. The results demonstrate our proposed perceptual loss can generate photo-realistic images without grid-like artifacts.

The objective of this study is to propose, implement, and evaluate a human-detection-based work time measurement method for a multi-product assembly and machining process with a small number of workers, where the work content and work location match and the work time and dwell time at the location also match. The system measures the work time based on the time spent in the work area when the Intersection over Union (IoU) between the pre-defined work area rectangle and the human detection rectangle exceeds a threshold value. A packaged system centered on a single-board computer was developed for on-site operation based on the basic concept of "anyone," "anywhere," and "easily". Trial experiment were conducted by the developers and actual operation experiment were conducted by manufacturing workers, on the constructed system. In the trial experiment, precision, recall, and accuracy all exceeded 90% as a result of the work time measurement. However, in the actual operation experiment, the results of work time measurement showed that precision and accuracy were above 90% and 80%, respectively, while recall was significantly lower, at 3.3% in the lowest case. This decrease in recall is due to the oversight of the work condition, which is mainly caused by the failure of human detection, and can be improved by changing the location of cameras or increasing the number of cameras. Overall, the results suggest the effectiveness of a work time measurement method based on human detection and a packaged system that implements the method.

We propose a method for recognizing the conversation activity in a group of pedestrians walking outdoors using a camera video sequence. Various existing methods have been developed for analyzing interaction in conversation, e.g., action recognition of whether people are talking or not. However, a method has yet to be proposed to recognize the conversation activity in a group of pedestrians. In this paper, we design a method to extract features for conversation activity recognition by using body interactions that occurred from pedestrians' conversations. Our method generates an interaction video sequence in a virtual road space using a temporal pose signal and temporal position signal representing body interactions. Our method determines a conversation activity label using the interaction video sequence, which is an informative and visible feature. Experimental results show that our method using an interaction video sequence can recognize the conversation activity more accurately than comparison methods using a temporal pose signal and temporal position signal.

The search activities to Great East Japan Earthquake are periodically done by many volunteers. Although their search area is extremely wide, the activities are done by human wave tactics. To make such activities more efficient, it is important that analysis of drone video for generating 3-D point cloud from a scene. If we can detect the different places between two point clouds obtained on different days, the detected places have high possibilities for missing objects. In this paper, we proposes a new method for aligning 3-D point clouds using dominant plane information. Our method uses a new keypoint-based feature descriptor for an almost planar surface point cloud and estimate a scale using the dominant plane in the point cloud. Our method can be applied to different scale point clouds. We show the effectiveness of our method by several experiments using simulated data from actually measured point clouds.

Ultrafine bubbles (UFBs) are microbubbles with a diameter of 1 μm or less. In recent years, many reports have shown that grinding performance and machining accuracy can be improved by using coolant containing UFB (UFB coolant) as the coolant in the grinding process. However, the mechanism by which this effect occurs has not been clarified. Existing studies consider that UFB coolant has better wettability than normal coolant, which changes grinding performance. If the inclusion of UFB changes the liquid properties of the coolant, then grinding with UFB coolant should be effective for any work material. However, there are cases where the effect of UFB is not obtained. Therefore, wettability is not considered to be the main cause of the UFB effect. As a general characteristic of UFB, it is known that OH radicals are generated when UFB collapses, which is thought to promote oxidation of the mating material. This study examines and discusses the mechanism of UFB effect generation in grinding from the viewpoint of oxidation of the work material surface. In the case of UFB water, the oxidation state of the work material surface was found to differ depending on the gas type. The composition of the oxides formed also differed depending on the gas type. Furthermore, friction and wear tests simulating the grinding process were conducted, and it was found that the friction coefficient varied depending on the type of gas in the UFB coolant, suggesting that the oxidation of the work material surface by UFB affects the friction and wear properties, resulting in a change in grinding properties.