Assistive Smart Cane for Visually Impaired People Based on Convolutional Neural Network (CNN)
Farhat Ullah, Ajantha Devi, Muhammad Abul Hassan, Iziz Ahmad, Muhammad Sohail, Muhammad Awais Mahoob, Hazrat Junaid · River Publishers eBooks · 2023
According to the World Health Organization, there are millions of visually impaired persons throughout the world who struggle to move freely. They always need help from people with normal sight. For visually impaired persons (VIP), finding their way to their desired destination in a new area is a huge difficulty. This research aimed to assist these individuals in resolving their challenges in moving to any place on their own. To this end, we proposed a method for VIP using a convolutional neural network (CNN) to recognize 126 the condition and scene objects automatically. The proposed system consists of Arduino UNO, ultrasonic sensors, a camera, breadboards, jumper wires, a buzzer, and an earphone. Breadboards are used to connect the sensors with the help of Arduino UNO and jumper wires. The sensors are used for the detection of obstacle and potholes while the camera performs as a virtual eye for the visually impaired people by recognizing these obstacles in any direction (i.e., front, left, and right). An important feature is provided by this system, in which the blind receives the scene object, the system automatically calculates how far he is away from the obstacles, and a voice message alerts him and directs him via earphone. The obtained experimental results show that the CNN yielded impressive results of 99.56% accuracy and has a loss validation of 0.0201%.