一句话：提出完全基于计算机视觉的移动应用自动化测试方法，从外部 GUI 信息推断功能并生成测试策略。
现有移动应用自动化测试通常依赖程序内部信息，例如读取代码、GUI 布局文件或捕获事件流，这限制了测试的交互性、灵活性与真实性。本文提出一种完全从视觉视角出发的自动化测试方法：借助计算机视觉技术，让计算机像人类一样仅根据移动应用的外部 GUI 信息判断其内部功能，并据此生成测试策略并执行。作者认为，这种基于视觉的测试途径有助于提升测试的交互性、灵活性与真实性，并缓解当前移动应用测试需求巨大而测试人员相对不足的矛盾。摘要仅阐述了方法思路与预期效果，未给出具体实现细节与实验数据。
The current implementation of automated mobile apps testing generally relies on internal program information, such as reading code or GUI layout files, capturing event streams. This paper proposes an approach of automated mobile apps testing from a completely visual perspective. It uses computer vision technology to enable computer to judge the internal functions from the external GUI information of mobile apps as we humans do and generates test strategy for execution, which improves the interactivity, flexibility, and authenticity of testing. We believe that this vision-based testing approach will further help alleviate the contradiction between the current huge test requirements of mobile apps and the relatively lack of testers.
DOI 原文 ·
@inproceedings{paperbot6039,
title = {Automated mobile apps testing from visual perspective},
author = {Feng Xue},
booktitle = {ISSTA},
year = {2020},
doi = {10.1145/3395363.3402644}
}