Webcam-Based Detection of Visual Ocular Impairments Based on Opacity, Shape, Ring Appearance and Position Using Media-Pipe Face Mesh For Accessibility-Aware Web Personalization

Abstract

Visual impairments caused by ocular abnormalities can directly affect a user's ability to perceive and interact with digital content. The standard eye examination normally relies on specialized imaging equipment and professional experience. This paper proposes a non-invasive, webcam-based framework. It detect visually observable ocular impairments and generate personalized accessible web-interface. The RGB video captured from a regular laptop webcam is processed with Media-Pipe Face Mesh, which is used to identify facial landmarks as well as eye and iris landmarks and pupils, before going through the visual analysis of the disease. The proposed  framework is based on four complementary approaches to detecting ocular abnormalities: position-based, opacity-based, shape-based, and appearance-based analysis using rings. The modules cover the following 7 target conditions: squint, cataract, sclerocornea, corneal scarring, corneal ulcer, iris coloboma and Arcus Senilis. Scale-normalized ocular measurements are employed to reduce sensitivity to variations in user-to-camera distance and head orientation. The extracted features are integrated into a unified impairment profile. Thisse profiles subsequently mapped to personalized accessibility adaptations such as text enlargement, contrast enhancement, magnification, spacing optimization, and simplified interaction. Evaluation on a 1,000-sample test dataset produced an overall classification accuracy of 98.20%. The model demonstrate the feasibility of the proposed approach for webcam-based ocular-condition screening. The framework also offers a means for integrating ophthalmic impairment assessment with adaptive human–computer interaction while eliminating the need for dedicated ophthalmic imaging equipment.