Offline Fraud Signature and Real Signature Classification Using Deep Convolutional Networks

EasyChair Preprint 8863

Abstract

Signature examination is an important area of expertise that has a legal character in human life. Advanced microscopes, lighting devices and signature verification methods are used in forensic and private document examination laboratories. Signature forgery has increased significantly in recent years. Our project aims to automate whether the signature is fake or real using deep convolutional neural networks. GPDC dataset is used. Accuracy rates were compared by using Alexnet, Resnet, Vgg16 architectures from deep learning CNN methods. It has been observed that the accuracy rate is higher in Alexnet architecture, which is one of these architectures used.

Keyphrases: AlexNet, CNN, Derin Öğrenme, ResNet, VGG16, İmza İnceleme

@booklet{EasyChair:8863,
  author    = {Tuba Talo and Ahmet Çınar},
  title     = {Offline Fraud Signature and Real Signature Classification Using Deep Convolutional Networks},
  howpublished = {EasyChair Preprint 8863},
  year      = {EasyChair, 2022}}