plagiarism detection on handwritten document images

plagiarism detection using crnn

1. Plagiarism Detection in Handwritten Document Images Submitted by: Bharat Gaurav MT19AIE226 M.Tech AI Guided by: Prof Anand Mishra, Assistant Professor, Computer Science IIT Jodhpur

2. Research Motivation Academic Integrity Prevent academic dishonesty by detecting similarities between handwritten documents. Matching Handwritten Assignments Forensics Determining the authenticity of handwritten documents is essential in legal proceedings. Mining old Documents Mining Healthcare documents Historical documents

3. The Problem Plagiarism Detection Word spotting Feature-based Deep learning models Hybrid methods Similarity score calculation

4. Challenges The variability of handwriting Handwriting can vary considerably between different people and even between the same person at different times. The complexity of handwritten text Handwritten text can contain a lot of noise such as smudges, ink bleed, and background clutter The difficulty of extracting features The features that can be used to identify similarities between handwritten documents are often subtle and difficult to extract.

5. Literature and SOTA OCR Feature based methods CNN CRNN CNN + LSTM Attention Open problems Document with Graphics Mathematical Equation Other scripts like Devnagari Semantics based plagiarism detection

6. Work Item Explore different deep learning architecture and loss function to improve performance Improving the transfer learning process to reduce the domain gap between synthetic and real domain Cover up the Dataset unavailability using data augmentation techniques Investigate the use of attention mechanism to improve model performance

7. Dataset The IAM Handwriting Database https://fki.tic.heia-fr.ch/databases/iam-handwriting-database iiit-hw https://github.com/kris314/hwnet/blob/master/iiit-hws/README.md

8. Project Progress Literature survey : Done Area of improvement : Done Research on improvement work : In Progress Implementation: Yet to start

9. References Matching Handwritten Document Praveen Krishnan and C.V Jawahar End to End System for Handwritten Text Recognition and Plagiarism Detection using CNN & BLSTM Gaurav Mukesh Shipurkar; Rishil Ripal Sheth; Tanish Ashok Surana; Kunal Nirav Shah; Rachit Garg; Prachi Natu A Robust Approach to Plagiarism Detection in Handwritten Documents A survey of document image word spotting techniques

10. CRNN Model for Handwritten Text Recognition The diagram below illustrates the CRNN model used for handwriting recognition.