Deep Learning based Intelligent Alert System for Visually Impaired People

According to a study of World Health Organization, there are about 39 million people in the world who are completely visually impaired. These individuals often face significant challenges in navigating and managing their daily life. They always need to be dependent on assistance from other people or any assistance system even for a short distance. So, the main objective of this work is to design a feedback system which acknowledges the person about the obstacles with their names. The frame in front of the user is given as the input to the system through a wearable device. Further, it detects the obstacles in that frame and the user is alerted with the name of the obstacle through a headphone or earphone connected. This can be integrated with a smart cane and any assistance system which improves the safety and mobility of the user. This is achieved through deep learning. Accuracy is the main cause for using deep learning. Better results can be attained by deep learning models. The system has shown 96% accuracy. The two main stages of the system, object detection & classification and the text to speech conversion are achieved through a deep learning technique and speech synthesizer respectively.