Visual Assistance for Blind Person by Using Machine Learning Technology

The advent of powerful computers and the availability of massive amounts of data has sparked renewed interest in Machine Learning. Many fields now make use of machine learning, from the processing of medical images to the development of fully driverless vehicles. Object detection in photographs is becoming a major field of study. Bounding boxes can now be drawn on detected objects by computers. Computer vision is another name for this. To help the visually impaired and the blind, we recommended using computer vision machine learning techniques for object detection. This project details the steps required to train a convolutional neural network on the ImageNet dataset so that it can perform object detection and provide descriptive narration for a visually impaired user. The ultrasonic waves emitted by these sensors are utilised for obstacle detection, while other types of sensors can tell you whether or not there's a fire nearby, how deep the water is, and whether or not it's day or night outside. When an impediment or a fire is detected, a buzzer alerts the person. When there is water or darkness ahead, a vibration motor alerts the traveller. The system places a call via GSM to the nearest caretakers if anything out of the ordinary happens. The suggested framework identifies an item in their immediate vicinity and conveys feedback in the form of dialogue; readers are alerted to these messages via headphones. The proposed framework is meant to provide a straightforward and fruitful path. The purpose of the deterrent discovery for the sight impaired is to let them to walk independently by giving them a sense of fake vision by providing data about the natural position of the static and dynamic elements around them.