This study proposes an automatic evaluation method for volleyball reception performance using video analysis techniques. The method analyzes volleyball matches recorded from the end line using computer vision and machine learning. First, the court and net are detected, and then the ball is tracked throughout the rally. Based on the ball trajectory, the system detects toss and spike events. The setter's position is identified through player detection, and the reception performance is automatically classified into three levels (A, B, C pass) according to the setter's movement distance. Experiments using volleyball match videos from YouTube achieved a court detection accuracy of 60%, ball tracking recall of 95%, toss and spike event detection recall of 85%, and reception performance evaluation accuracy of 63%. The proposed method provides objective evaluation of reception performance, which can assist coaches and players in improving techniques and strategies. Future work includes enhancing detection accuracy and introducing skeleton-based action recognition for more comprehensive analysis.

This paper proposes an automated method for identifying over-the-table stroke types in table tennis using image processing. In table tennis, stroke types such as high-speed type (chiquita, flick), stop type (stop), and deep return type (push, flow, etc.) performed near the table are crucial for gaining advantage, yet their analysis is time-consuming and often subjective. Our proposed method employs YOLOv8 to detect players, table, and ball from standard match footage, and then identifies bounce points to distinguish over-the-table techniques from other strokes. The identification algorithm analyzes ball trajectory, bounce positions, and stroke speed to categorize stroke types, with machine learning methods achieving superior performance. Additionally, the method classifies forehand and backhand drives by analyzing player posture at the moment of striking. Experiments on World Table Tennis match videos showed high accuracy: 99.6% for table detection, 96.1% for ball detection, and 89.9% for bounce point detection. The identification of over-the-table stroke types achieved 85.1% accuracy using machine learning methods, while forehand and backhand drives were correctly classified at 90.2% accuracy using rule-based methods. This automated system provides coaches and players with an objective and efficient tool for analyzing technical performance in table tennis.

This paper proposes a method for automatically estimating the number of rallies in table tennis match videos as part of improving the efficiency of play analysis. By using this method, the manual effort previously required for recording rally counts can be eliminated. The analysis assumes a vertically oriented camera view, as commonly seen in broadcast footage. Since the ball in the lower court may be occluded by players, only the upper court is analyzed. The ball in the upper court is detected by analyzing the pixel value differences between the ball and the table. The number of rallies is measured by counting the number of times a ball hit by a player passes through the upper court. To ensure accurate counting, hit balls are distinguished from high-toss serves and failures due to the net. Experimental results show that the proposed method correctly counted the number of rallies in 87.0 % of 623 scenes from Japanese T-League match videos. Furthermore, when the existing scoring/losing rate calculation method was applied to the automatically measured rally counts, the results closely matched those obtained from manually recorded rally counts, confirming the effectiveness of the proposed rally counting method. Large-scale quantitative analysis becomes feasible through automated processing.

This study presents an automatic method for rally scene detection and score estimation in a broadcast table tennis video. Pre-match analysis typically requires extensive manual recording of player actions, shot trajectories, and scores from videos. This time-intensive process impacts preparation efficiency. The proposed methodology integrates YOLOv8 for player and table detection with Segment Anything Model (SAM) for scoreboard area segmentation. Rally scenes are identified through pixel change rate analysis combined with Mean Square Error calculations for frame similarity assessment. The research introduces a score recognition technique functioning independently of scoreboard design variations. Experimental evaluation demonstrates a 9 percentage point increase in precision compared to existing methods, with reduced false positives and missed detections. By incorporating score change timing data, the system overcomes challenges presented by variable camera angles and scene transitions. This work contributes to automated sports video analysis by demonstrating the feasibility of detection systems that enhance pre-match preparation while reducing manual analytical workload in professional table tennis contexts.

Smart cities are being promoted due to the declining population. In smart cities, AI using sensing information such as camera image information provides various services. However, for financial reasons, it is important to realize various types of AI with limited use (high-mix low-volume) at low-cost as AI in smart cities. In terms of the cost of AI generation, it is necessary to consider the design cost and the training cost. In addition, high resistance to adversarial examples (adversarial robustness) is also important for practical applications of AI. Adversarial training is reported to be effective in achieving high adversarial robustness. Therefore, this study proposes a low-cost high-mix low-volume edge AI as a smart city-oriented AI. The proposed edge AI is based on VAT, which combines semi-supervised learning and AT to reduce learning cost. The proposed edge AI consists of a binary neural network (BNN) and uses high-level synthesis for its implementation to reduce design cost. This reduces design and training costs and creates a secure edge AI that can handle a small volume and variety of products.

As the demand for wireless communications continues to grow, securing limited spectrum resources has become an issue. Under such circumstances, Dynamic Spectrum Allocation (DSA), in which existing frequency bands are jointly used and allocated to carriers, has been attracting attention. However, the problem is that a large number of base stations and interferers cause a combinatorial explosion, which requires an enormous amount of computation time. Saito et al. proposed a method to solve this combinatorial optimization problem using quantum annealing, and succeeded in significantly reducing the computation time ⁽¹⁾. However, they focused only on the frequency requirements divided by each time axis and did not consider the requirements regarding the length of the time axis. Furthermore, considering bursty streams is important in DSA since internet traffic includes bursty streams. In this paper, we focus on the DSA problem of maximizing the satisfaction of time-based requests from each base station including bursty traffic and prioritized sessions, and describe the formulation for solving the problem using quantum annealing. Experimental validation using capable annealing machines confirms the efficacy of the proposed formulation in optimizing base station traffic demands within DSA. These results validate the feasibility of quantum annealing for DSA and provide valuable insights for further advancements in spectrum management.

The IEEE 802.15.4 beacon mode has been proposed as a wireless standard for controlling devices and systems. Its superframe consists of Contention Access Period (CAP) based on CSMA/CA and Contention Free Period (CFP) based on TDMA. Therefore, the controller needs to perform communication scheduling to determine whether to communicate with each plant to be controlled and, if so, which communication segment, CAP or CFP, to use. A Previous study proposed a joint optimization method for communication scheduling and control inputs for each controlled object based on Model Predictive Control (MPC). However, the conventional method involves a combinatorial optimization problem, leading to an exponential increase in computational complexity with respect to the number of control targets, slots, and prediction periods. In this study, we propose a sequential optimization method that increases the computational complexity only linearly with respect to these factors. Simulation results demonstrate that the effectiveness of the proposed method in reducing the computational complexity while also evaluating its performance degradation.

This paper proposes a framework for compensating fluctuations of renewable energy generation by adjusting charging amount of Electric Vehicles (EV) in a microgrid. We have three purposes: matching supply and demand, minimizing the microgrid's cost, and maintaining the convenience of EV while adjusting charging amount depending on the generation from renewable sources. In order to achieve these purposes, we propose two-stage framework, Stackelberg-game approach, and considering expected utility of EV. As a result, the microgrids could determine their energy management strategy while considering their cost and EV's convenience, and therefore positive economic effect has been created, matching supply and demand.

We have worked on a community system design introducing the Gift & Circulation Model of relationality assets to activate communication among residents by providing gift-giving opportunities and lowering participation hurdles. The primary objective of this research is to alleviate residents’ isolation, and it aims to introduce a system-driven, proactive mechanism that enhances incentives for voluntary participation in the community system and fosters new connections and relationships among individuals. To investigate the possibility of creating new communities and strengthening existing ones, the system divides residents into groups and reimburses them with points that increase when they give gifts to groups, as determined by the system. In addition, to investigate what residents respond best to the assets involved, the upper limit of reimbursement points to be given back and the points themselves are changed to reflect the results of community activities or to compensate for the loss of one's assets. The field experiment results showed that it was effective in strengthening interactions within existing communities. We also found that assigning meaning to reimbursement points, such as a reward for community activities or compensation for the loss of assets, stimulates interaction among residents more than if they were not given such a meaning.

GRODMIC (GROup Delay between MICrophones) is a method proposed as a high spatial resolution source direction estimation for two microphones. Since a spatial aliasing is not required to consider for its applying, spatial resolution can be improved by extending the microphone width. In this method, the propagation delay characteristics between microphones are estimated by an approximation model, and the TDOA (Time-Difference-Of-Arrival) is estimated from the group delay characteristics. Therefore, the accuracy of group delay characteristics is important. However, multiple ripples tend to occur in the estimated group delay characteristics due to the high order model together with increasing of the wide width microphone pair. Due to this degradation of estimation accuracy, the advantage of high spatial resolution is lost. As a result, high direction estimation performance is limited to a situation to be able to apply a low order model. In this study, a reduction method of model order is proposed based on the approximation model characteristics.

In this paper, we propose a power lines detection method for automatic power lines removal. When detecting power lines in an image, the difficulty of detection depends on the features of the power lines background. Therefore, in this paper, four wire detection networks are trained by focusing on the differences in background regions obtained by semantic segmentation technique. The proposed method achieves wire detection by combining four wire maps obtained from four networks. Experiment results show the effectiveness of the proposed method.

Differential Evolution with a selection mechanism based on the probability of mutation strategies for constrained optimization is proposed in this letter. The proposed method is based on the principles of diversification and intensification, and adjusts the selection probability based on the Gompertz curve in a scheduled manner. Numerical experiments demonstrate this method outperforms both single-strategy approaches and those employing equal probability selection.