Enhancement of Web Accessibility for Persons with Disability PwD Using Machine Learning

Abstract

In the present scenario the primary activity for Visually Impaired (VI) individuals that creates opportunities for social media connection as part of professional, political, and social life is web access. Online image access and comprehension are two challenges VI users encounter. The development of assistive technology based on computer vision (CV) helps persons with visual impairments (VI) in a variety of situations, including grocery shopping, creating alternate text, recognising objects, comprehending text documents, and recognising people, among others. An automated system is created to generate alternative (alt) text for online images that are not captioned or for which the alt text is not defined in order to make the digital platform user-friendly for VI people. A practical approach for VI that enables automatic captioning of web image is provided by the Deep Belief Network - Bald Eagle Search (DBN-BES) method. Choosing photographs without captions is the first step, and the Bald Eagle Search (BES) Algorithm is used to make this decision. After the selection phase, the Deep Belief Network (DBN) model is used to generate alt text for the associated photos newline