Gender Recognition Method Based on Convolutional Neural Networks
Cai Shiwe · Video Engineering · 2014
When using artificial intelligence for gender recognition,the face would be influenced by illumination and obstructions,those face make this work more difficult. In this paper,a class of convolutional neural networks for gender classification is proposed. These networks are built upon the new structure,which enhance the network's classification ability. In order to solve the optimal rejection area of the sample,the confidence measure are given. Experiments show that the performance of the gender recognition has been well improved,the method achieve 98. 67% correct recognition rate by rejecting 7. 46% test sample.