Persian Signature Verification using Fully Convolutional Networks

TL;DR

This paper introduces the use of Fully Convolutional Networks (FCNs) for Persian signature verification, demonstrating improved accuracy over traditional CNNs by leveraging raw signature images and global pooling.

Contribution

The paper proposes a novel application of FCNs with global average pooling for signature verification, replacing conventional CNNs and enhancing feature extraction from raw images.

Findings

FCN with global average pooling outperforms CNN on UTSig database.

FCN effectively learns robust features directly from raw signature images.

The approach simplifies the network architecture while improving accuracy.

Abstract

Fully convolutional networks (FCNs) have been recently used for feature extraction and classification in image and speech recognition, where their inputs have been raw signal or other complicated features. Persian signature verification is done using conventional convolutional neural networks (CNNs). In this paper, we propose to use FCN for learning a robust feature extraction from the raw signature images. FCN can be considered as a variant of CNN where its fully connected layers are replaced with a global pooling layer. In the proposed manner, FCN inputs are raw signature images and convolution filter size is fixed. Recognition accuracy on UTSig database, shows that FCN with a global average pooling outperforms CNN.