Project Overview

This project focuses on developing an intelligent AI system that can automatically detect handwritten signatures from scanned document images and compare them across different documents to determine their similarity. The goal is to enable efficient and accurate signature verification by analyzing critical visual features such as shape, stroke flow, alignment, and texture.

Manual signature comparison is time-consuming, error-prone, and subjective. By leveraging computer vision and deep learning techniques, this project aims to automate the process while maintaining high precision and reliability. 

Objectives

• Build an AI system to detect handwritten signatures from images of documents.
• Compare two or more signatures to determine similarity and authenticity.
• Provide an interpretable similarity score that reflects how closely two signatures match.

Expected Outcome

• A working prototype that detects and extracts signatures from diverse document types (PDFs, scanned images, photos, etc.).
• A comparison model capable of computing a similarity score between signatures with an accuracy target above 90%.
• A simple interface or API for users to upload two document images and receive a similarity report.
• Foundational dataset and benchmark metrics for future research in handwritten signature verification.

How to Apply

We invite candidates and research teams to submit their project proposal for this open problem. Your proposal should include:

• Literature Review: Summarize relevant work and techniques in signature detection and verification, highlighting any models, datasets, or metrics previously used.
• Proposed Methodology: Describe the techniques, models, or algorithms you plan to use (e.g., CNN-based feature extraction, Siamese networks, Vision Transformers, or hybrid methods).
• Implementation Plan: Outline the workflow, data requirements, preprocessing steps, training strategy, and evaluation metrics.
• Innovation or Differentiation: Explain what makes your approach unique or more effective than existing methods.
• Deliverables and Timeline: Specify key milestones such as dataset preparation, model training, evaluation, and final report submission.