Vision sense: Object Detection and Navigation Aid for the Visually Impaired

Abstract—living in a world without vision is fraught with many challenges that sighted people take for granted. For the visually impaired, even mundane activities like walking through crowded areas, detecting obstacles, or identifying objects can be intimidating and even hazardous. Important views are like white canes and training dogs, and information provision limitations. With recent changes in computer vision and machine learning, there is an increasing chance to create smart systems that can improve accessibility to blind people. VisionSense is a new object detection and navigation system’s goal is to overcome these challenges by combining computer vision technologies. By using real-time image processing and deep learning based object detection algorithms, VisionSense is able to detect and inform the user about the presence and location of objects within their surroundings. Coupled with initiative audio or haptic feedback mechanisms, the system enables users to make informed navigation choices and steer clear of possible dangers. This work describes the design, development of Vision Sense, emphasizing its capability to greatly enhance the quality of life for the visually impaired. The system is not just affordable and light weight but also flexible to indoor and outdoor spaces, making it a viable remedy for accessible, real-time wayfinding aids. Keywords—Python, Flask, OpenCV, YOLO (YOU ONLY LOOK ONCE) v8, HTML, CSS, Java stack, Bootstrap, ML(MACHINE LEARNING).