ECSS Paris 2023: CP-AP12
INTRODUCTION: Physical fitness encompasses the bodys capacity to adjust comprehensively. More workforce and material resources must be put into the fitness assessment, which can lead to inconsistent data quality due to different examiners or assistants. The data collected from the assessment tends to be fragmented or singular, needing continuous process records and analysis of dynamic motion. With the advancement of science and technology and the growing trend of sports technology, scientific data is collected and measured conveniently through innovative computer vision technology. METHODS: This study aims to establish the necessary technological framework for motion capture and body tracking to analyze the landmark data of human bodies in a video and to develop computer vision technology to design and develop the physical fitness detection system for college students. This research designed an image data collection environment, developed a computer vision-based physical fitness test system using Googles MediaPipe framework to measure and validate data of physical fitness testing items, and enhanced Physical Fitness Testing procedures. RESULTS: The dimensions of the experimental space are 300 centimeters in length, 380 centimeters in width, and 250 centimeters in height. We analyzed data from 6 participants. The data collection sessions were conducted individually and were stored for later analysis. The preliminary result shows that the sit-ups, sit and reach, and standing long jump is high as large, more than 0.9; the correlation via logistic regression models suggests that the relationship between the traditional method and the computer vision-based technology beliefs is very strong. CONCLUSION: AI imaging technology brings benefits to physical fitness testing. The system can capture human body motion via pose detection and pose tracking. The studys overall architecture is feasible and practical.
Read CV PEI-CHING YANGECSS Paris 2023: CP-AP12
INTRODUCTION: With the rise of sports technology and Artificial Intelligence (AI), the integration of sensors with AI is gaining popularity in the sports industry. The emerging technology of Tiny Machine Learning (TinyML), developed in recent years to address the limited computational resources of devices, is being widely adopted. This study introduces a cost-effective The Real-time Table Tennis Swing Classification System Based on TinyML. The system aims to reduce the training burden on table tennis coaches or players and enhance the interest of beginners in learning professional techniques. METHODS: The system is primarily comprised of two main components: the hardware, known as the embedded AI smart table tennis racket, and the software, referred to as the computer application. The hardware aspect, the embedded AI smart table tennis racket, includes the racket itself and an RF wireless receiver for receiving measurement signals. The software component, the Computer Application, is responsible for displaying the classification results. This combination establishes a real-time table tennis swing posture classification system based on TinyML technology. After collecting the swing data, it is transmitted to the AI-MCU (Microcontroller Unit) on the Embedded AI Smart Table Tennis Racket. The AI-MCU uses its TinyML model for real-time swing posture classification. The classified swing data is then wirelessly transmitted via Radio-Frequency (RF) to the computer-side application for display. The embedded AI smart table tennis racket utilizes the STM32F7 component as the main control chip. The STM32F7 component is responsible for communication and control of 6-axis sensorts and performs processing and calculations for the TinyML model. RESULTS: In the data analysis, real-time classification is achieved using AI deep learning theory. By collecting swing data from our schools players, four different AI deep learning models are tested for data segmentation and classification accuracy of swing data. The experimental results show that Convolutional Neural Networks (CNN) have the best capability for waveform segmentation and swing classification accuracy. In practical swing tests, the converged training data and optimized CNN model achieve an average classification accuracy of approximately 98.3% for actual table tennis player swings. Through the development of the Embedded AI Smart Table Tennis Racket, it not only enables swing posture classification but also corrects the swing postures of typical table tennis beginners, achieving the intelligent training goal for table tennis. CONCLUSION: Through the development of the Real-time Table Tennis Swing Classification System Based on TinyML, it not only enables swing posture classification but also corrects the swing postures of typical table tennis beginners, achieving the intelligent training goal for table tennis. In the future, this research can also be combined with AR/VR related applications.
Read CV Yung-Hoh SheuECSS Paris 2023: CP-AP12
INTRODUCTION: Advances in running shoe technology focus on combining embedded carbon fiber plates with novel midsole foams to improve running performance. Variations in midsole stiffness and energy return (ER) across brands have been shown to influence running economy (RE) [1,2]. Differences in shoe last, foam composition, plate stiffness, and other specific design features make it difficult to attribute the importance and interaction between midsole stiffness and ER. The purpose of this study was to assess changes in RE in response to different midsole mechanical properties within a singular brand’s line. METHODS: This study examined three shoes of a singular brand’s line (M1, M2, M3) (Mass = 218, 210, 210g respectively [3.7% difference]), heel stack (42, 36, 33mm respectively [27.3% difference]) featuring nitrogen infused EVA foam midsoles. Only M1 contained a carbon fiber plate. Stack height was measured at the heel with sock liner intact under a load of 5N. The shoes were uniaxially loaded over 100 cycles (687N; 70kg body mass [BM] equivalent). Force was applied separately at the heel and midfoot regions, and stiffness and ER were obtained. 8 trained male runners (Age: 22.3 +/- 2.3yr, BM: 63.9 +/- 2.0kg, World Athletics Score: 904.9 +/- 159.0 pts) of the same shoe size (US9) completed a 10-min warm-up (12.39kph) followed by two, 3-min trials in each shoe (14.86kph). Shoes were tested in a randomized and mirrored order. Blood lactate was measured at rest, and after trials 1, 4, and 6 to ensure participants were below aerobic threshold (defined as an increase in blood lactate >1.5mmol/L above resting). RE was calculated using a 5 breath moving average of VO2 from the last minute of each trial and averaged between both trials run in the same shoe per subject. A repeated-measures ANOVA was used to assess differences between heel and midfoot stiffness and RE. RESULTS: Shoes differed in stiffness at the heel (M1=63.3; M2=64.3; M3=79.1 kN/m [25% max difference]) and midfoot (M1=54.4; M2=81.4; M3=72.0 kN/m [49.6% max difference]). ER was similar across all three models at the heel (M1=73.1; M2=72.8; M3=72.6% [.7% max difference]) and midfoot (M1=74.4; M2=72.2; M3=73.0% [3% max difference]). RE was not significantly different (p=0.057) between shoes, which differed up to 1.61% (M1=186.3 +/-13.1ml/kg/km, M2=188.3 +/- 10.9ml/kg/km, M3=189.3+/-11.9ml/kg/km). CONCLUSION: In the present sample of trained men of similar body mass, RE did not significantly differ in shoes whose midsole foams differed in stiffness but not ER. A more diverse subject pool and greater variations in ER between shoes may be needed to identify the magnitude of ER necessary for significant improvements in RE. Maximizing ER may be more impactful than altering midsole stiffness when determining foam properties that maximize distance running performance. REFERENCES: 1. Burns et al, 2023 ECSS. 2. Joubert et al, 2023 Int J Sport Phys Perf.
Read CV Danielle PageECSS Paris 2023: CP-AP12