SUPPORTING PATIENTS OF AGE-RELATED MACULAR DEGENERATION IN MALAYSIA USING MOBILE APPLICATION ENABLED WITH OBJECT AND TEXT RECOGNITION

Abstract

When people are diagnosed with Age-related Macular Degeneration (AMD) disease, object or text recognition through their eyesight pose a challenging task as they are not blind but lose the central vision. Despite various assistive technologies available to help AMD patients to see better in advanced or western countries but there is lack of research how lightweight mobile application with object and text recognition capability to help Malaysian AMD patients especially in situations without Internet connection. The research objective is to understand Malaysian AMD patient’s requirement then develop a simple but useful mobile application with object and text recognition capability to generate audio speech for patients to perceive and understand their surrounding environment better. After literature reviewed, appropriate machine learning models are selected and trained for object and text recognition functionality. Then the mobile application is developed using Kotlin programming language and Android Studio before series of testing are conducted. Three AMD patients also are selected to test the mobile application. The developed mobile application yields mean Average Precision (mAP) of 42% for object recognition and 83% for text recognition. Despite lower mAP for object recognition however through Unit and Usability Testing conducted on AMD patients, the test results met their requirements and the mobile application did support them to have a better grasp of their surroundings. This research contributed a small step in relieving Malaysian AMD patients to see better and live more independently. Limitations and future improvements were also provided.