| Summary: | 碩士 === 國立臺灣科技大學 === 電子工程系 === 105 === This paper proposes an algorithm that can trace swimming algorithm, such as stroke counts, laps and swimming styles in real time. Our algorithm consists of two distinct parts. The first part is sensory data processing phase and the second part is stroke analysis phase. In the beginning of our algorithm, sensory data processing phase collects the data from triaxial accelerometer and eliminate floating signal. After that, our algorithm will judge whether the user starts to swim or not. Next part, stroke analysis phase, which computes stroke counts, laps and swimming styles.
The experiment revealed that the average accuracy of left and right hand with our algorithm in stroke counts is 97.04%, in lap counts is 96.73% and in swimming style recognition is 92.69%. The average accuracy of left and right hand with commercially available swim watch (Garmin) in stroke counts is 92.15%, in lap counts is 88.91% and in swimming style recognition is 82.24%. According to the experimental results, the algorithm, proposed by this paper, has a higher accuracy than other devices and other algorithms. Therefore, our algorithm can efficiently compute stroke counts, laps and swimming styles in real time.
According to the experimental results, the algorithm, proposed by this paper, has a higher accuracy than other devices and other algorithms. Therefore, our algorithm can efficiently compute stroke counts, laps and swimming styles in real time.
|
|---|
