Abstract:

Blind people encounter many difficulties while engaging with their surroundings. Therefore, this project proposes a deep learning method that assists the blind people in an indoor environment. It detects the presence of objects, regardless of their position. This process can be called as image multi-labeling. In our project, the system will capture a picture of the situation present in front of the person and then provide information about it. The processed image will be voice-delivered to the user along with any objects and text in it. The suggested method recommends employing a web camera to identify items in real-time video using object detection. The You Look Only Once (YOLO) model, a real-time object detection method based on CNN, is used. Also, the software program and deep learning method are implemented using the Python OpenCV modules. The Google text-to-speech library is used to deliver image recognition results to blind people in the form of audio and to pinpoint an object's location in relation to its position on the screen

Description:

Object detection, real time, OpenCV, YOLO, deep learning, Visually Impaired.