Personalized Interaction Techniques of Vision-Based 3D Dynamic Gestures Based on Small Sample Learning
Huiyu Wu · Dianzi xuebao · 2013
There are some unresolved issues left behind for many traditional dynamic gesture recognition methods,such as Hidden Markov Model(HMM),Neural Network(NN),and statistical classifiers.For example,they require a large number of training examples and the involvement of expert users in the training process.Moreover,they are used for some specific gesture sets which are difficult to be extended.In this paper,we first build a task model and a state transition model for vision-based dynamic gestures.Then we propose a method for 3D dynamic gesture recognition based on small sample learning.Next we design a toolkit for development of user-defined gestures.Finally,we develop a gesture-based interactive television prototype.Experimental results verify the validity of our method.