Deep Learning Enabled Novel Blind Assistance System for Enhanced Accessibility

Developing a smart blind assistance system utilizing deep learning techniques to enhance accessibility for blind people. This study focuses on two groups. Group 1 uses YOLOv3 which is an object detection algorithm used to detect objects using image dataset. Group 2 Open CV is an object detection algorithm used to detect objects using a video dataset for object detection but lacks action detection. Group 3 employs the Convolutional Neural Network (CNN) algorithm for real-time action detection with improved accuracy. The Convolutional Neural Network (CNN) algorithm did really well at spotting actions as they happened, scoring an average of 81% accuracy. By adding voice commands to VGG16, blind assistance systems become more accessible and provide a better user experience than the YOLOv3 and Open CV framework.