Utilizing HTML‐analysis and computer vision on a corpus of website screenshots to investigate design developments on the web

We present preliminary results of a project investigating the design development of popular websites between 1996 and 2020 via HTML analysis and basic computer vision methods. We acquired a corpus of website screenshots of the current top 47 popular websites. We crawled a snapshot of every month of these websites via the wayback machine of the Internet Archive platform since the time snapshots are stored to gather 7,953 screenshots and HTML pages. We report upon quantitative analysis results concerning HTML elements, color distributions and visual complexity throughout the years.