We have trained the siamese network for the face recoginition.
I have used the dataset of university students in which there are 20 photos of each grad student and total number of students are 150. I have trained our siamese network on those images and tested out network on other images.
Firstly, I have prepared the data for training by forming pairs of images which are alike and also pair which are not and assign them target as 1 and 0 respetively. Then dataset is splited into training and test set for training and testing we got the accuracy of 97% on traning set and 96.7 of accuracy on test set.
Also, I have used images different from the dataset on which i have trained the network and got the accuracy of 75%. But it can improve by further training the deep convolution neural network. We can improve further performance of network by increasing number of epochs and improve the picture quality to extract more features from the images.