ACTION RECOGNITION WITH VOICE OUTPUT USING DEEP LEARNING FOR ASSISTING BLIND PEOPLE

ACTION RECOGNITION WITH VOICE OUTPUT USING DEEP LEARNING FOR ASSISTING BLIND PEOPLE

Prof. S. K. Sabnis1, Abhishek Manwatkar2, Janvi Patil2 , Bhakti Sutar2, Rahul Wankhade2

1 Assistant Professor of the Department, Department of Information Technology, Rajiv Gandhi Institute of Technology

2Students, Department of Information Technology, Rajiv Gandhi Institute of Technology

---------------------------------------------------------------------***---------------------------------------------------------------------

Abstract - In an era marked by technological advancements, accessibility remains a critical concern, particularly for individuals with visual impairments. "Action Recognition Using Deep Learning for Blind People" addresses this issue by proposing a comprehensive framework that leverages deep learning techniques to assist blind individuals to understand their surroundings. This project encompasses various modules which aimed at providing real-time voice output to visually impaired individuals.

"Action Recognition Using Deep Learning For Blind People" is a groundbreaking project focusing at enhancing convenience and independence for visually impaired people through innovative technology solutions. This project focuses on developing a comprehensive system that integrates various modules, including sign language detection, object detection, text extraction from PDF documents, and real-time text extraction. Leveraging TensorFlow Lite for image preprocessing and OCR engines for text extraction, the system generates audio output to convey information effectively to blind users.

Key Words:  Action Recognition, Deep Learning, TensorFlow Lite, Image preprocessing, OCR engines, Text extraction