NiCro: Purely Vision-based, Non-intrusive Cross-Device and Cross-Platform GUI Testing

TL;DR

NiCro is a non-intrusive, vision-based system that enables cross-device and cross-platform GUI testing by accurately matching GUI widgets across diverse devices without intrusive system modifications.

Contribution

It introduces a novel, purely vision-based approach for GUI widget matching that works across different devices and platforms, avoiding intrusive system modifications.

Findings

NiCro outperforms four existing matching techniques in widget and GUI matching accuracy.

It successfully recorded and replayed 107 test cases across 8 devices, demonstrating robustness.

The system effectively supports cross-platform GUI testing without intrusive instrumentation.

Abstract

To ensure app compatibility and smoothness of user experience across diverse devices and platforms, developers have to perform cross-device, cross-platform testing of their apps, which is laborious. There comes a recently increasing trend of using a record and replay approach to facilitate the testing process. However, the graphic user interface (GUI) of an app running on different devices and platforms differs dramatically. This complicates the record and replay process as the presence, appearance and layout of the GUI widgets in the recording phase and replaying phase can be inconsistent. Existing techniques resort to instrumenting into the underlying system to obtain the app metadata for widget identification and matching between various devices. But such intrusive practices are limited by the accessibility and accuracy of the metadata on different platforms. On the other hand, several recent works attempt to derive the GUI information by analyzing the GUI image. Nevertheless, their performance is curbed by the applied preliminary visual approaches and the failure to consider the divergence of the same GUI displayed on different devices. To address the challenge, we propose a non-intrusive cross-device and cross-platform system NiCro. NiCro utilizes the state-of-the-art GUI widget detector to detect widgets from GUI images and then analyses a set of comprehensive information to match the widgets across diverse devices. At the system level, NiCro can interact with a virtual device farm and a robotic arm system to perform cross-device, cross-platform testing non-intrusively. We first evaluated NiCro by comparing its multi-modal widget and GUI matching approach with 4 commonly used matching techniques. Then, we further examined its overall performance on 8 various devices, using it to record and replay 107 test cases of 28 popular apps and the home page to show its effectiveness.