Application of Unsupervised Deep Learning in Color Image Recognition
Download as PDF
Li Bingjie, Gao Jingyang, Shi Xinyu
Today is the Internet information age, now deep learning has made a breakthrough in the field of image, and convolution neural network is the leader of this wave of deep learning wave. In reality, images are not in the environment of simple background and single target. It is difficult to recognize such images by using CNN model. In this paper, the application of unsupervised deep learning in color image recognition is proposed. In order to verify the performance of SVM image recognition method, this paper takes two training samples and test samples, and selects different hidden layer nodes and layers to test. Through analysis, it is concluded that the accuracy of two hidden layers is higher than that of only one hidden layer. In order to further test the feasibility of the algorithm, in cooperation with a University of traditional Chinese medicine, typical tongue images were selected as training samples. Experiments are carried out on facial expression database and tongue image data respectively. The recognition rate of all images can reach more than 92%, and the recognition time is about 43M / s. The results show that the algorithm is in the international high level, and the image recognition is more intelligent. Through the analysis, the research in this paper has achieved ideal results, and made a contribution to the application of unsupervised deep learning in color image recognition method.
Deep Learning, Convolution Neural Network, Perceived Loss, Image Recognition