-
Notifications
You must be signed in to change notification settings - Fork 2
Gesture recognition
CAGE’s current gesture recognition system uses the uWave gesture recognition system designed by Jiayang Liu, Zhen Whang and Lin Zhong of Rice Univerity, and Jehan Wickramasuriya and Venu Vasudevan of the Pervasive Platforms and Architectures Lab at Motorola Labs. Their system employs Dynamic Time Warping to match gestures created with different speeds, and is specifically designed for accelerometer-based gesture recognition.
You can find a white paper on their recognition system at: http://www.ruf.rice.edu/~mobile/publications/liu08uwave.pdf
CAGE’s former gesture recognition system used the $1 Gesture recognizer, designed by Jacob Wobbrock and Yang Li of the University of Washington, and Andrew Wilson of Microsoft Research. depts.washington.edu/aimgroup/proj/dollar/ is the homepage, which contains a JavaScript implementation of the recognizer, as well as links to a paper and pseudocode about the $1 gesture recognizer.
Some other options we’re considering:
- Gesture recognition based on SVMs (Support Vector Machines): www.springerlink.com/content/k1q27m3918711318/fulltext.pdf
- Gesture recognition using Dynamic Time Warping and Hidden Markov Models: www.springerlink.com/content/k1q27m3918711318/