DefinePK

Abstract

Visually impaired individuals experience significant mobility challenges due to the limited situational awareness provided by traditional aids like white canes. To address this, an AI-powered smart vision kit that enhances environmental perception through real-time object detection and audio feedback has been proposed. Our system combines embedded edge computing with optimized neural networks to deliver a portable, low-cost assistive solution. The hardware prototype incorporates a Raspberry Pi 4 and camera module with a TensorFlow Lite pipeline, utilizing a quantized MobileNetV1 SSD model trained on the COCO dataset for efficient inference. The framework processes live video streams via OpenCV, detecting objects within a 5-meter range at 12 FPS (tested on 480p input). Detections are converted to spatialized audio alerts using text-to-speech (TTS), prioritizing critical obstacles.