WebAccessVL: Making an Accessible Web via Violation-Conditioned VLM
This work addresses web accessibility for users with disabilities by automating violation corrections, though it is incremental as it builds on existing VLM and program synthesis methods.
The paper tackles the problem of automatically editing website HTML to address Web Content Accessibility Guidelines 2 (WCAG2) violations using a vision-language model, resulting in a reduction of average violations from 5.34 to 0.44 per website and outperforming commercial LLM APIs.
We present a vision-language model (VLM) that automatically edits website HTML to address Web Content Accessibility Guidelines 2 (WCAG2) violations. We formulate this as a supervised image-conditioned program synthesis task, where the model learns to correct HTML given the HTML and its rendering. We collected WebAccessVL, a new dataset with manually corrected accessibility violations, establishing paired training data. We then propose a violation-conditioned VLM that additionally conditions on the WCAG2 violation count to guide the correction process. Experiments demonstrate that our method effectively reduces the average number of violations from 5.34 to 0.44 per website, outperforming commercial LLM APIs (Gemini, GPT-5). A perceptual study confirms that our edited websites maintain the original visual appearance and content.