A Gesture Recognition Research Based on Unsupervised Feature Learning
-
Abstract
In order to solve static gesture image classification problems, the paper presents a unsupervised learning process combining the static supervised classification.According to the unsupervised sparse auto-encoder neural network, the image patches were trained to extract the edge feature of the image, and these edge features are the input of a classifier, and finally.And the classifier parameter was used to improve the classification accuracy.
-
-