Automated human vision assessment using computer vision and artificial intelligence

This paper presents an automated system to assess human vision to identify early signs of vision disorders such as amblyopia (lazy eye), so that potential problems can be addressed as early as possible by having the system refer children to a specialist (pediatric ophthalmologist). The system does not require extensive operator training or patient cooperation. This paper explores the application of photoscreening, computer vision and artificial intelligence techniques for diagnosing vision disorders by processing video images taken of patientspsila eyes, computing important eye features, and determining the referral decisions. Extensive experiments and analysis indicate that the system has an accuracy of 77% when evaluated using the referral decisions, which are recommended by a specialist.