<![CDATA[Access to visual information and social interaction remain significant challenges for individuals with visual impairments, particularly in recognizing surrounding environments and identifying nearby people independently. Existing assistive applications generally focus only on object detection or require human assistance or additional hardware, limiting their practicality and support for user independence. This study aims to develop AEyes, an Android-based assistive application integrating MediaPipe and FaceNet for facial recognition and a Large Language Model (LLM) for contextual environmental description with Text-to-Speech features. The research employed a Research and Development (R&D) approach using the Waterfall model. Application feasibility testing was conducted based on ISO/IEC 25010 standards. Data were collected using system validation instruments for functional suitability, usability questionnaires for user satisfaction, and observation for portability, performance efficiency, maintainability, and usability effectiveness and efficiency. Usability testing involved 10 students at SLB-A Yapti Makassar. Results indicate that the application meets high technical feasibility and usability criteria. Functional Suitability achieved a score of 1, usability effectiveness reached 96%, efficiency was 0.04488 goals/sec, and user satisfaction reached 86% (very satisfied). The application demonstrated efficient performance with average CPU usage of 5.6% and memory consumption of 251.3 MB. It also achieved 100% portability across 6 different Android versions and a Maintainability Index of 99.89 (Excellent). These findings indicate that AEyes is feasible for use and has the potential to support user independence based on initial testing results. However, this study is limited by its small-scale testing and has not yet evaluated long-term impacts in real-world usage.]]>
