Scholarlink

<p>Navigating complex environments poses a significant challenge for visually impaired individuals who often rely on traditional aids such as guide dogs and white canes. This aids in the lack of real-time, precise spatial feedback. This paper proposes an advanced place recognition system that utilizes YOLO-based object detection and 3D audio feedback to enhance spatial awareness. The system provides a portable camera and deep-learning algorithms to detect nearby objects and generate spatialized audio cues that guide users safely. The results show an improvement in object recognition accuracy, with the Enhanced YOLO with Attention model achieving 91% accuracy, surpassing the baseline YOLO model by 10%. Additionally, the proposed system demonstrates robustness in occluded environments, with only a 6% performance drop under 30% occlusion compared to 15.5% in the baseline model. User evaluations indicated improved usability, with an easy-to-use rating of 4.8 and Feedback Clarity of 4.7 on a 5-point scale, supporting the effectiveness of the 3D audio approach for independent navigation. These results suggest that the proposed system provides significant enhancement in accessibility, allowing visually impaired individuals to navigate with greater confidence.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Navigating beyond sight: a real-time 3D audio-enhanced object detection system for empowering visually impaired spatial awareness

  • 期刊论文

  • 发表日期：

    2025年9月5日

查看全文
  • Ankit Kumar,
  • Abhishek kumar,
  • Rohit Raja,
  • Amit Kumar Dewangan,
  • Manoj Kumar,
  • Aradhana Soni,
  • Dheeraj Agarwal

摘要

Navigating complex environments poses a significant challenge for visually impaired individuals who often rely on traditional aids such as guide dogs and white canes. This aids in the lack of real-time, precise spatial feedback. This paper proposes an advanced place recognition system that utilizes YOLO-based object detection and 3D audio feedback to enhance spatial awareness. The system provides a portable camera and deep-learning algorithms to detect nearby objects and generate spatialized audio cues that guide users safely. The results show an improvement in object recognition accuracy, with the Enhanced YOLO with Attention model achieving 91% accuracy, surpassing the baseline YOLO model by 10%. Additionally, the proposed system demonstrates robustness in occluded environments, with only a 6% performance drop under 30% occlusion compared to 15.5% in the baseline model. User evaluations indicated improved usability, with an easy-to-use rating of 4.8 and Feedback Clarity of 4.7 on a 5-point scale, supporting the effectiveness of the 3D audio approach for independent navigation. These results suggest that the proposed system provides significant enhancement in accessibility, allowing visually impaired individuals to navigate with greater confidence.