This is implementation of Siamese Neural Network for counting people based on their face image. Network architecture that i used in this project was implemented from: https://github.com/Goldesel23/Siamese-Networks-for-One-Shot-Learning with a few modification. This network architecture was provided to extract face features. Then, the extracted features will be used to measure the distance between two compared images.

Beta v1.0

1. New Siamese Architecture
2. New Dataset (Datatrain)
3. New Function for Counting
4. New Function for Recognition

I'm using new datatrain here. This datatrain obtained from: Face94 by Dr Libor Spacek (http://cswww.essex.ac.uk/mv/allfaces/faces94.html). This dataset contains 3080 images in total from 153 different people. This images will be used as datatrain for training process. Each data will have a pair of images and a labels (1.0 for same person, 0.0 for different person). dataset.py generate permutation 2 in every person (class) for data with same person (label 1.0), and then each images will be paired with 10 different images from 10 different person (label 0.0)