Gesture Recognition using Hidden Markov Models
This project presents an approach for recognizing and classifying different arm motion gestures by using IMU sensor readings from gyroscopes and accelerometers to train a set of Hidden Markov Models, which gives the result as log-likelihood for each class and the class with the highest likelihood is chosen as the result.
The data format of the IMU readings collected from consumer mobile device is:
ts | Ax Ay Az | Wx Wy Wz |
---|---|---|
Time (ms) | 3x Accelerometer (m/s2) | 3x Gyroscope (rad/s) |