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

Area of Science:

• Computer Science
• Artificial Intelligence
• Assistive Technology

Background:

• Current visual aid systems face limitations in conventional algorithms, sensor capabilities, power consumption, and cloud reliance, leading to latency and privacy concerns.
• Existing assistive technologies for the visually impaired lack secure communication and struggle with complex computer vision tasks.

Purpose of the Study:

• To introduce SENSEYE, a resource-aware visionary framework utilizing edge computing and secure communication for enhanced assistive technology.
• To develop a portable, cost-effective, and power-efficient system integrating IoT Edge and virtual decentralized services.

Main Methods:

• Integration of open AI models (e.g., SSD-MobileNetV2, VILA1.5-3b) for real-time object detection, obstacle localization, and scene comprehension.
• Implementation of edge computing for low-latency processing, reducing reliance on cloud services.
• Incorporation of secure communication mechanisms for remote assistance, video streaming, and GPS location sharing.

Main Results:

• SENSEYE provides accurate environmental awareness and auditory feedback through optimized lightweight AI models and efficient speech processing.
• The system demonstrates superior accuracy, power efficiency, and responsiveness compared to traditional sensor-based or cloud-reliant systems.
• Enabled secure remote assistance and real-time GPS location sharing, enhancing user safety and connectivity.

Conclusions:

• SENSEYE presents a novel resource-aware edge AI architecture for next-generation assistive systems for the visually impaired.
• The framework establishes a scalable, privacy-preserving, and interpretable foundation for enhanced accessibility.
• Future work will focus on improving wearability, energy optimization, and multimodal AI integration for user-centric accessibility.