A gestural human computer interface for smart health

For centuries, man was forced to live a highly active lifestyle with food being a precious commodity. Technological advances in the past few decades have resulted in increasingly sedentary lifestyles and a surfeit of calorie dense foods. This has resulted in a global epidemic of obesity and a host of associated health problems. One way to address this problem is to incorporate a higher level of physical activity into the workday. The objective of this thesis is to design a low cost gestural human computer interface for the recognition of vigorous gestures. We demonstrate that an action vocabulary of eight intuitive gestures can be recognized by the use of inexpensive accelerometers and a computationally simple approach involving Principal Component Analysis and Naïve Bayes classification. The accuracy is comparable to more computationally intensive approaches. The actions can be mapped to commands for controlling commonly used applications like e-mail and customized to individual preferences. There is a significant rise in pulse rate during these actions comparable to light aerobic activity. This has the potential to mitigate the harmful effects of sedentary work habits by raising the rate of metabolism with minimal impact on productivity.