WebUI: A Dataset for Enhancing Visual UI Understanding with Web Semantics

TL;DR

This paper introduces WebUI, a large dataset of 400,000 web pages with metadata, to improve visual UI understanding, especially in mobile applications, by leveraging web semantics despite noisy data.

Contribution

The paper presents WebUI, a large-scale web-crawled dataset with automatically extracted metadata, and demonstrates strategies to enhance visual UI understanding models using web semantics.

Findings

WebUI contains 400,000 web pages with metadata.

Incorporating web semantics improves UI understanding in mobile.

Strategies like element detection and screen classification boost model performance.

Abstract

Modeling user interfaces (UIs) from visual information allows systems to make inferences about the functionality and semantics needed to support use cases in accessibility, app automation, and testing. Current datasets for training machine learning models are limited in size due to the costly and time-consuming process of manually collecting and annotating UIs. We crawled the web to construct WebUI, a large dataset of 400,000 rendered web pages associated with automatically extracted metadata. We analyze the composition of WebUI and show that while automatically extracted data is noisy, most examples meet basic criteria for visual UI modeling. We applied several strategies for incorporating semantics found in web pages to increase the performance of visual UI understanding models in the mobile domain, where less labeled data is available: (i) element detection, (ii) screen classification and (iii) screen similarity.