Scholarlink

<p>Designing and creating a real-time assistive system to help visually impaired people safely and independently navigate indoor environments is the main goal of this project. The foundation of this effort is a thorough analysis of existing assistive technologies, which identified significant shortcomings in terms of object recognition precision, scalability, and adaptability in confined or cluttered environments. With an emphasis on both single-stage and two-stage architectures, the study systematically evaluates deep learning-based object detection techniques in order to address these issues. Building on these discoveries, a hybrid model is put forth that combines a quantum-enhanced classifier with YOLOv8, a top single-stage detector, in order to increase classification accuracy and dependability. The suggested model performs well on the COCO dataset, achieving 86.6% training accuracy and 83.3% test accuracy. Notably, the model’s capacity to detect objects of different sizes and complexity is improved by the quantum classifier, which improves detection performance by honing ambiguous or low-confidence outputs.</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

Comparative performance analysis of single-stage and two-stage object detectors and design of novel quantum-enhanced deep learning model for visually impaired

  • 期刊论文

  • 发表日期：

    2025年10月4日

查看全文
  • B. N Rashmi,
  • R. Guru,
  • M. A Anusuya,
  • Mallikarjuna Korrapati

摘要

Designing and creating a real-time assistive system to help visually impaired people safely and independently navigate indoor environments is the main goal of this project. The foundation of this effort is a thorough analysis of existing assistive technologies, which identified significant shortcomings in terms of object recognition precision, scalability, and adaptability in confined or cluttered environments. With an emphasis on both single-stage and two-stage architectures, the study systematically evaluates deep learning-based object detection techniques in order to address these issues. Building on these discoveries, a hybrid model is put forth that combines a quantum-enhanced classifier with YOLOv8, a top single-stage detector, in order to increase classification accuracy and dependability. The suggested model performs well on the COCO dataset, achieving 86.6% training accuracy and 83.3% test accuracy. Notably, the model’s capacity to detect objects of different sizes and complexity is improved by the quantum classifier, which improves detection performance by honing ambiguous or low-confidence outputs.