loading page

Framework for Face recognition and Scene Description using Deep Learning for Visually Challenged people
  • S. Suseela,
  • Felicia Sharon
S. Suseela
Vellore Institute of Technology - Chennai Campus

Corresponding Author:[email protected]

Author Profile
Felicia Sharon
Vellore Institute of Technology - Chennai Campus
Author Profile

Abstract

Independent living is becoming more and more important in today’s society, where persons with visual impairments struggle with a lot of limitations. Visually impaired people are at a disadvantage since they need manual help to compel information about their surroundings. Since many tasks require visual information, persons who are visually impaired find it difficult to complete many of life’s essential chores, such as going for a stroll, eating, and conversing with a group of people. The development of technology has made it feasible to assist the blind. With the advent of recent advances in Artificial Intelligence Techniques and Deep Learning models, this project proposes a sophisticated system to help visually challenged people. The system uses Deep Learning Faster R-CNN ResNet-50 FPN for object detection, CNN ResNet-50 for Face recognition, and CNN ResNet-50 and LSTM for Image Captioning. Then a voice message is generated to assist the Visually Impaired person.