Developing signature recognition model using CNN architecture. Transfer learning from VGG16 network is used to extract features of dataset in initial layering. Pre trained weights of VGG16 is used to predict features and then these fatures are fed into fully connected layers and output layer is softmax layer which gives classification score on two class, i.e class-1 forged, class-2 genuine.
akinuzelgencer / signature-biometrics Goto Github PK
View Code? Open in Web Editor NEWThis project forked from adilfaiz001/signature-biometrics
Signature Biometrics - using transfer learning.