J. ArninD. AnopasP. TriponyuwasinT. Yamsa-ArdY. WongsawatMahidol University2018-11-092018-11-092014-02-122014 Asia-Pacific Signal and Information Processing Association Annual Summit and Conference, APSIPA 2014. (2014)2-s2.0-84949924564https://repository.li.mahidol.ac.th/handle/20.500.14594/33680© 2014 Asia-Pacific Signal and Information Processing Ass. Exercise is a good alternative approach to be healthy. However, it can cause a body in negative outcome for people who over workout without proper manner. Therefore, the objective of this project is to develop an activity tracker called 'Feelfit' that has a high accuracy to measure levels of activity (with 5 intensities of exercises). Besides, challenging and motivating the exercise via feedback of detailed information such as burned calorie, activity percentage, and so on are proposed. An accelerometer, high accuracy tri-axial accelerometer (MMA8452Q), sends a value of acceleration in 3 axes acquired from body movement and then be processed and calculated physical activity behavior on a low-power microcontroller. This project has proposed a novel algorithm of physical activity classification and calories burned calculation. The algorithms were examined the accurateness by 10 healthy subjects (5 males and 5 females) aged 15-25 years old. The proposed algorithms were also compared with a commercial activity monitoring device; the accuracy of calories burned calculation is more than 80% and more than 90% for activity classification.Mahidol UniversityComputer ScienceConference PaperSCOPUS10.1109/APSIPA.2014.7041798