Scholarlink

<p>Despite significant recent advances, visual aid systems are still limited by the use of conventional computer vision algorithms, constrained sensor capabilities, high power consumption, and reliance on cloud-based processing, which introduce latency and privacy risks. Current assistive technologies for visually impaired individuals often suffer from a lack of secure and competent communication and an inability to deal with complex computer vision tasks. This paper introduces SENSEYE, a resource-aware visionary framework that employs edge computing with a secure and competent communication mechanism. The proposed architecture integrates IoT Edge and virtual decentralized services in a portable system that is small, cost-effective, and power-efficient. The integration of open AI models with advanced functionality in this system design helps to recognize objects, locate moving obstacles, detect sudden changes, perceive a summary from a live feed, and represent them as audio in real time. SENSEYE integrates real-time object detection, scene comprehension, and global positioning system (GPS)-based navigation into a portable, low-latency device. This system leverages optimized lightweight AI models, e.g., SSD-MobileNetV2 and VILA1.5-3b, to provide accurate environmental awareness and seamless auditory feedback through efficient speech processing. The system also enables secure remote assistance via video streaming and real-time GPS location sharing, ensuring enhanced user safety and connectivity. The evaluations confirm superior accuracy, power efficiency, and responsiveness compared to traditional sensor-based or cloud-reliant systems. Although challenges remain, particularly in constrained sensor capabilities, power efficiency trade-offs, and potential sensory overload, future work will focus on improving wearability, optimizing energy consumption, and advancing multimodal, user-centric AI integration for enhanced accessibility. This paper presents a novel resource-aware edge AI architecture with integrated real-time perception and secure IoT communication, establishing a scalable, privacy-preserving, and interpretable foundation for next-generation assistive systems for visually impaired individuals.</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

SENSEYE: a resource-aware visionary framework for assisting individuals with visual disabilities

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月11日

查看全文
  • Akash Shingha Bappy,
  • Tapio Seppänen,
  • Md Ziaul Hoque

摘要

Despite significant recent advances, visual aid systems are still limited by the use of conventional computer vision algorithms, constrained sensor capabilities, high power consumption, and reliance on cloud-based processing, which introduce latency and privacy risks. Current assistive technologies for visually impaired individuals often suffer from a lack of secure and competent communication and an inability to deal with complex computer vision tasks. This paper introduces SENSEYE, a resource-aware visionary framework that employs edge computing with a secure and competent communication mechanism. The proposed architecture integrates IoT Edge and virtual decentralized services in a portable system that is small, cost-effective, and power-efficient. The integration of open AI models with advanced functionality in this system design helps to recognize objects, locate moving obstacles, detect sudden changes, perceive a summary from a live feed, and represent them as audio in real time. SENSEYE integrates real-time object detection, scene comprehension, and global positioning system (GPS)-based navigation into a portable, low-latency device. This system leverages optimized lightweight AI models, e.g., SSD-MobileNetV2 and VILA1.5-3b, to provide accurate environmental awareness and seamless auditory feedback through efficient speech processing. The system also enables secure remote assistance via video streaming and real-time GPS location sharing, ensuring enhanced user safety and connectivity. The evaluations confirm superior accuracy, power efficiency, and responsiveness compared to traditional sensor-based or cloud-reliant systems. Although challenges remain, particularly in constrained sensor capabilities, power efficiency trade-offs, and potential sensory overload, future work will focus on improving wearability, optimizing energy consumption, and advancing multimodal, user-centric AI integration for enhanced accessibility. This paper presents a novel resource-aware edge AI architecture with integrated real-time perception and secure IoT communication, establishing a scalable, privacy-preserving, and interpretable foundation for next-generation assistive systems for visually impaired individuals.