Handwritten Signature Verification System using Deep Learning

Sudharshan, Duth P.; Vismaya, R N

doi:10.1109/icdsis55133.2022.9915833

Cited by 12 publications

(3 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The experimental results indicate that verification accuracy increases when featurebased classifiers are combined. The model was built to validate signatures through the transfer of learning and activation functions for the three different CNN models (VGG16, VGG19, and ResNet50) with the addition of some parameters for each model, training, and testing on SigComp2009 dataset, showing that the VGG16 model has a high efficiency of 97%-compared to other models this approach was proposed by [28]. Thus, this study focuses on developing a system for offline signature verification using a CNN with a genetic algorithm that can search for the best model architecture hyperparameters.…”

Section: Related Workmentioning

confidence: 99%

Developing a Model for Offline Signature Verification Using CNN Architectures and Genetic Algorithm

Musleh,

Al-Azzani

2023

jast

View full text Add to dashboard Cite

The signature verification process has many applications, such as its use in financial operations, providing the electronic signature of documents, and providing an additional confidentiality standard to verify the identity of users in computer systems. This process has the advantage of being accepted by the community and is less intrusive compared to other biological methods. Deep learning (DL) and CNN (Convolutional Neural Network) are widely used by bioinformaticians. Due to the difficulty in extracting features in other systems or models, DL and CNN-based signature verification systems have been significantly improved. Yet Hyperparameter optimization for CNN models remains a challenging problem in designing highly efficient models with the most accurate results. It is often convolutional neural network (CNN) models that are manually designed. The proposed method is focused on a genetic algorithm that develops a population of CNN models in order to find the best fit architecture for designing an offline signature verification model. Our model is tested on more than one dataset, BHSig260-Bengali, BHSig260-Hindiin, GPDS, and CEDAR. The result of the approach proposed in this paper has the highest discrimination rate of FRR of 2.5, FAR 3.2, EER 2.35, and 97.73 %-accuracy rate.

show abstract

Section: Related Workmentioning

confidence: 99%

Developing a Model for Offline Signature Verification Using CNN Architectures and Genetic Algorithm

Musleh,

Al-Azzani

2023

jast

View full text Add to dashboard Cite

show abstract

“…Signature verification [1] is a fundamental modality for legally and socially confirming an individual's identity on a global scale. A reliable and robust signature verification system plays a vital role in sectors such as banking, finance, security, and legal documentation, serving as a means to detect and prevent fraud and forgery.…”

Section: Introdctionmentioning

confidence: 99%

Signature Verification Based on Deep Learning

Salama

2023

Alexandria Journal of Science and Technology

View full text Add to dashboard Cite

Signature verification is considered one of the main features in determining the person identity. Our proposed framework emphasizes the potential of Deep Learning Models (DLMs) in revolutionizing signature verification techniques and underscores the need for continuous exploration and advancement in the realm of automated signature authentication. Therefore, five pre-trained DLMs, ResNet50, DenseNet121, MobileNetV3, InceptionV3, and VGG16, based on four different datasets, CEDAR, BH-Sig260 Bengali, BHSig260 Hindi, and ICDAR 2011(Dutch),are introduced in this paper to verify the person identity. Furthermore, data augmentation techniques are applied to overcome dataset limitations and increase the framework's performance. Additionally, transfer learning and finetuning techniques are performed to reduce computational time and memory usage. It is observed that the InceptionV3 DLM based on the ICDAR 2011 (Dutch) achieved the best performance of 100% accuracy, 100% AUC and 100% sensitivity. While, CEDAR Dataset achieves performance with an accuracy of 99.76%, an AUC of 99.94%, sensitivity of 99.76%, precision of 99.76%, an F1-score of 99.71%, score, and a computational time of 13.627s.

show abstract

“…Reference [20] utilized deep convolutional networks to learn representations of image features for image retrieval. In signature offline verification, many works use CNNs to learn representations of images [21][22][23][24]. In the seal recognition and verification tasks, the method of using CNNs to automatically extract seal features [15] has a higher accuracy.…”

Section: Introductionmentioning

confidence: 99%

Representation Learning Method for Circular Seal Based on Modified MLP-Mixer

Cao,

Zhou,

Zhang

et al. 2023

Entropy

View full text Add to dashboard Cite

This study proposes Stamp-MLP, an enhanced seal impression representation learning technique based on MLP-Mixer. Instead of using the patch linear mapping preprocessing method, this technique uses circular seal remapping, which reserves the seals’ underlying pixel-level information. In the proposed Stamp-MLP, the average pooling is replaced by a global pooling of attention to extract the information more comprehensively. There were three classification tasks in our proposed method: categorizing the seal surface, identifying the product type, and distinguishing individual seals. The three tasks shared an identical dataset comprising 81 seals, encompassing 16 distinct seal surfaces, with each surface featuring six diverse product types. The experiment results showed that, in comparison to MLP-Mixer, VGG16, and ResNet50, the proposed Stamp-MLP achieved the highest classification accuracy (89.61%) in seal surface classification tasks with fewer training samples. Meanwhile, Stamp-MLP outperformed the others with accuracy rates of 90.68% and 91.96% in the product type and seal impression classification tasks, respectively. Moreover, Stamp-MLP had the fewest model parameters (2.67 M).

show abstract

Handwritten Signature Verification System using Deep Learning

Cited by 12 publications

References 10 publications

Developing a Model for Offline Signature Verification Using CNN Architectures and Genetic Algorithm

Developing a Model for Offline Signature Verification Using CNN Architectures and Genetic Algorithm

Signature Verification Based on Deep Learning

Representation Learning Method for Circular Seal Based on Modified MLP-Mixer

Contact Info

Product

Resources

About