Gender Classification Through Face Recognition Using MLP Neural Network
Shahani Imam Bux 12MSIT08 Supervisor - Dr. Sajida Parveen Soomro
Gender Classification Through Face Recognition Using MLP Neural Network - Nawabshah QUEST 2019 - 29p
ABSTRACT
Artificial Neural Networks are playing vital role in machine learning. ANNs are also used for different fields of artificial intelligence. But it is most su itable for pattern recognition. Gender classification is an also prom ising area of Al i n which ANNs plays good role. ANNs also called Multi-Layer Perceptron (MLP), wh ich is widely used for pattern classification, recognition. Here MLP is used for gender classification with back propagation algorithm. MLP shows good resu lts when basic gender face image features are represented to it. This thesis work proposed a methodology for gender classification through face recognition by using MLP neu ral network. Training of MLP with male and female face images, the output are also categorized according to input images. For gender classification, face images should
be scaled out at same dimensions and same color. By using different dimensions of
face images with RGB color mode, the MLP gives good results up to 90%. Results are also altered when MLP attributes are changed. But it is tested that by using one hidden layer with 12 neurons generates good results. Some results are simulated through Neuroph Studio, a java based platform for neural network simulation.
Gender Classification Through Face Recognition Using MLP Neural Network - Nawabshah QUEST 2019 - 29p
ABSTRACT
Artificial Neural Networks are playing vital role in machine learning. ANNs are also used for different fields of artificial intelligence. But it is most su itable for pattern recognition. Gender classification is an also prom ising area of Al i n which ANNs plays good role. ANNs also called Multi-Layer Perceptron (MLP), wh ich is widely used for pattern classification, recognition. Here MLP is used for gender classification with back propagation algorithm. MLP shows good resu lts when basic gender face image features are represented to it. This thesis work proposed a methodology for gender classification through face recognition by using MLP neu ral network. Training of MLP with male and female face images, the output are also categorized according to input images. For gender classification, face images should
be scaled out at same dimensions and same color. By using different dimensions of
face images with RGB color mode, the MLP gives good results up to 90%. Results are also altered when MLP attributes are changed. But it is tested that by using one hidden layer with 12 neurons generates good results. Some results are simulated through Neuroph Studio, a java based platform for neural network simulation.