Design and Implementation of a Cross Platform i0s Application Development Framework Based on YAI Configuration Files

This study focuses on addressing the travel challenges faced by visually impaired individuals, particularly in China where there are approximately 17 million affected. A cross-platform iOS application development framework based on YAI has been designed and implemented, leveraging smartphone technology and integrating computer vision and deep learning for 3D object detection. The framework optimizes neural networks for fast and energy-efficient detection, incorporating a comprehensive process from data acquisition, preprocessing, scene segmentation, object detection to 3D modeling. Experimental results demonstrate high accuracy in stair detection (95%) and indoor object detection (exceeding 80% on iOS devices and over 90% in augmented reality). This framework not only enhances object detection accuracy and efficiency but also provides convenient and safe auxiliary tools for visually impaired individuals, supporting the entire development cycle from data collection to final object detection. Looking ahead, this study aims to expand its research scope to outdoor environments and further improve 3D detection accuracy to contribute to the development of computer vision and provide more accurate, convenient, and secure assistive tools for visually impaired individuals.