Classification of Gait Pattern with Wearable Sensing Data

Either instinctively or due to specific actions, human gaits involve the many ways in which people can move. The categorization of gait pattern has been researched in recent years to investigate and understand human behaviors using various kinds of data such as image data, floor sensing data and wearable sensing data. The performance of wearable sensors on smartphones for gait classification was studied in this research. Further, ensemble learning-based methods were presented to enhance the accuracy of gait pattern classification. Besides, their roles in two different positions on human body, comprising the waist position and the arm position, were studied for sorting three distinct gait patterns: walking upstairs, walking downstairs and walking on floor. From experimentation, ensemble learning-based approaches could enhance classification performance when using wearable sensors with an accuracy of 98.5333 percent, as shown by the experimental results.