With the growing popularity of electronic devices, Graphical User Interfaces (GUIs) are becoming increasingly important. Thus, GUI testing is also prevailing in software testing. Various approaches have been applied to automated GUI testing to guarantee software quality. However, almost all the existing approaches are platform-dependent due to their reliance on platform frameworks. These frameworks provide interfaces where testing tools get detailed information about the application under test, which results in lack of universality and efficiency.

To address the aforementioned challenges, we propose a novel software testing tool, named UniRLTest. UniRLTest leverages Computer Vision (CV) technologies and image understanding to achieve platform-independent usage. It extract all widgets in the given screenshot and construct a widget tree, where page information could be further inferred from.