Doctor’s Handwriting Detection & Text Summa...

Doctor’s Handwriting Detection & Text Summarization: Simplifying Prescriptions with Handwriting-to-Text Integration

Author Name : Chaitanya Mittapalli, Harika Vundru, Dr. V. S. Saranya

ABSTRACT The project "Doctor’s Handwriting Detection & Text Summarization" focuses on addressing the challenge of deciphering illegible handwriting, particularly from doctors, which can often lead to miscommunication in healthcare settings. Leveraging the Google Vision Pro API, a model named DTrOCR, specialized in recognizing handwritten characters, has been trained on the IAM Handwriting Dataset. To enhance its accuracy, the model has been fine-tuned with an additional 100-layer Convolutional Neural Network (CNN) and a 28-layer Hidden Markov Model (HMM). These enhancements result in an impressive accuracy of 98-99% in detecting and interpreting handwritten symbols and texts. The solution is embedded in a Streamlit application that allows for real-time handwriting detection as well as image uploads for analysis. By accurately transcribing and summarizing handwritten content, this project aims to streamline processes, reduce errors, and improve the efficiency of data handling in various sectors, particularly in healthcare.