Offline Signature Verification System Using Siamese Networks with Four Distinct Backbone Architectures
Our research paper published in IEEE ICECTE 2026 presenting an offline signature forgery detection system evaluated on the CEDAR dataset using MobileNetV2, VGG16, InceptionV3, and Custom CNN backbones.
I am proud to share our latest research work published in the 2026 5th International Conference on Electrical, Computer & Telecommunication Engineering (ICECTE), IEEE on January 31, 2026.
Paper Overview
Title: Offline Signature Verification System Using Siamese Networks with Four Distinct Backbone Architectures: MobileNetV2, VGG16, InceptionV3 and a Custom CNN Conference: IEEE | 2026 5th International Conference on Electrical, Computer & Telecommunication Engineering (ICECTE) Publication Date: January 31, 2026 DOI: 10.1109/ICECTE69292.2026.11429465 Keywords: Signature Verification, Siamese Networks, Deep Learning, Biometric Authentication, CNN, MobileNetV2, VGG16, InceptionV3
Abstract & Research Summary
This research presents an offline signature verification system based on Siamese Neural Networks using four distinct backbone architectures: MobileNetV2, VGG16, InceptionV3, and a Custom CNN. The study evaluates the effectiveness of deep learning techniques for biometric authentication and signature forgery detection using the standard CEDAR signature dataset.
- Architecture: Comparative evaluation of feature extraction capabilities across standard pre-trained architectures vs. custom lightweight CNN designs.
- Verification Metric: Contrastive loss evaluation to measure similarity distance between genuine and forged signature pairs.
- Application: Enhanced security solutions for banking, legal document processing, and automated verification workflows.
Publication Day & Presentation Experience
Presenting our paper at ICECTE 2026 was a rewarding milestone. Engaging with domain experts, discussing neural network optimization techniques, and receiving positive feedback from the computer vision research community reinforced my commitment to pursuing practical AI & deep learning research.