AI VISION AID: YOLO OBSTACLE DETECTION AND CLASSIFICATION

Abstract

This work presents a real-time obstacle detection and classification system designed to assist visually impaired individuals using Raspberry Pi and YOLO object detection. An ultrasonic sensor first detects the presence of an obstacle, triggering the camera to capture the scene, which is then processed by a lightweight YOLO model to identify and classify the object. The detected object label along with its position (left, right, or center) is converted into speech to inform the user. Additional safety features include an accelerometer to detect falls, which automatically sends an email alert with GPS location to a caretaker, and a panic button for manual emergency assistance. A moisture sensor identifies water-filled potholes, and an LDR determines whether it is day or night, enabling context-aware alerts. This multifunctional system enhances mobility and personal safety for the visually impaired in real-world environments.