Automated Visual Testing Workflow for Set-Top Box Issue Reproduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current visual testing methods for software or application functionality, particularly in content distribution platforms, are inefficient and prone to inconsistencies due to manual tagging and varied reporting styles, making it difficult to reproduce issues across different testers and devices.
Innovation Solution
A visual testing system utilizing a trained machine learning model to detect screen elements and automate test scenarios, allowing for efficient and consistent issue reproduction across various devices by generating workflows that can be shared and executed on different STBs or compatible devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual tagging and description of displayed features by testers is used, then testing can be performed independently of underlying design or code, but the process becomes lengthy and repetitive with poor issue reproduction capability
Solution Approach 1:
The patent replaces manual mechanical processes (testers physically observing and describing features) with an automated computer vision system using machine learning models that automatically detect and describe displayed features, thereby maintaining testing independence while dramatically improving efficiency
Solution Approach 2:
The patent introduces an intermediary automated testing system that acts as a mediator between the software under test and human developers. This system captures screenshots, uses machine learning to identify UI elements, and generates standardized issue reports, eliminating the need for direct human observation while preserving testing capability
2Ease of operation
If manual tagging by testers is used, then testing can be performed, but discrepancy between tester description and actual issue occurrence makes reproduction difficult or impossible
Solution Approach 1:
The patent replaces human testers' subjective descriptions with an automated system that objectively identifies and describes UI elements using machine learning. The system captures precise coordinates, element types, and text content, eliminating the discrepancy between description and actual occurrence that plagues manual testing
Solution Approach 2:
The patent creates accurate digital copies of the actual issue state by capturing screenshots and using machine learning to identify and describe the exact UI elements involved. This digital reproduction includes precise location data and element properties, enabling exact reproduction of issues without relying on imperfect human memory or description
3Adaptability or versatility
If varied reporting styles by different testers are used, then individual testing can be performed, but inconsistency makes issue reproduction across testers problematic
Solution Approach 1:
The patent enforces homogeneity in reporting by implementing a standardized issue report format generated automatically by the system. All issues are reported with consistent structure including screenshot data, identified UI element properties, location coordinates, and standardized descriptions, eliminating the variability introduced by different human testers' personal styles
Data Source
AI summary
An example method for visual testing and issue communication of programmed display of content includes obtaining a workflow of test scenarios for visual testing a display. The content displayed on the display is controlled by a set-top box (STB) device executing target instructions. The example method further includes identifying a visual testing issue associated with executing the test scenarios, and communicating the workflow and issue to a remote device for reproduction of the issue.