Fingerprint-recognition using deeplearning method I used siamese network to judge whether two fingerprint pictures are from same finger.
Reference《Learning a similarity metric discriminatively, with application to face verification》and《 Hamming Distance Metric Learning》
This project uses a siamese network that uses weight sharing and a contrastive loss function to learn a model which can judge whether two fingerprint pictures are from same finger.
1.Download fingerprint picture NIST has a lot of fingerprint pictures. https://www.nist.gov/itl/iad/image-group/resources/biometric-special-databases-and-software For example: Special Database 4 - NIST 8-bit Gray Scale Images of Fingerprint Image Groups.
2.Fingerprint process(optional) Reference code: https://github.com/Rubens10010/Fingerprint-Classification-System
3.Data augmentation The python scripts are in Data_augmentation directory.
4.Caffe install Reference my another github project: Caffe_Learning
5.Siamese network The network files are in Siamese_caffe directory
6.Training siamese network
7.Visualize the learned Siamese embedding
8.Test Test whether two fingerprints are the same.
9.Attach to raspberry pi Reference my another project: