Signature Recognition System Using ML
1st Prof. Vijet Swadi
Department of Computer Science and
Engineering(AI&ML)
KLS Vishwanathrao Deshpande
Institute of Technology Haliyal, India vbs@klsvdit.edu.in
4th Mr.Shivaprasad Kallappagoudar
Department of Computer Science and
Engineering(AI&ML)
KLS Vishwanathrao Deshpande
Institute of Technology Haliyal, India 2vd22ci048@klsvdit.edu.in
2nd Mr. Shivakumar C Chikkanaragund
Department of Computer Science and
Engineering(AI&ML)
KLS Vishwanathrao Deshpande
Institute of Technology Haliyal, India
2vd22ci047@klsvdit.edu.in
5th Mr. Mohammadyusuf R Basrekatti
Department of Computer Science and
Engineering(AI&ML)
KLS Vishwanathrao Deshpande
Institute of Technology Haliyal, India 2vd22ci022@klsvdit.edu.in
3rd Mr. Mahmadhujefa S Kattimani
Department of Computer Science and
Engineering(AI&ML)
KLS Vishwanathrao Deshpande
Institute of Technology Haliyal, India 2vd22ci020@klsvdit.edu.in
Abstract— This paper presents a machine learning-based system for offline signature verification, aimed at distinguishing between genuine and forged handwritten signatures. The proposed system utilizes a Siamese neural network architecture to extract and compare feature embeddings from two input signature images: a reference (genuine) signature and a query (test) signature. By computing the Euclidean distance between the feature vectors, the model determines the degree of similarity between the two signatures. A predefined threshold value of 0.2 is employed to classify the query signature as either genuine or forged. The system is developed using Python, leveraging TensorFlow for model training and inference, and Streamlit to provide a user-friendly web-based interface for real-time signature verification. Experimental results demonstrate the model's effectiveness in achieving reliable verification performance, highlighting its potential applications in domains such as document authentication, financial verification. transactions, and identity Varification.
Keywords : Signature verification, Siamese neural network, Machine learning, Offline handwriting recognition, Biometric authentication, Euclidean distance, Deep learning, TensorFlow, Image similarity, Streamlit.
