Autoscreenshot Verification for Virtual OS Integrity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automated screenshot verification systems struggle to accurately evaluate the integrity of virtual operating system states and backup images due to the ever-changing representations of bootable states across different operating systems, making it difficult to extract text from screenshots and match against dynamic lists of representations.
Innovation Solution
The method involves using a trained artificial neural network model to classify the state of screenshots taken during the boot process or execution of a process-of-interest, eliminating the need for text extraction and allowing for consistent evaluation of integrity by analyzing visual patterns.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If text extraction and matching against dynamic lists of representations is used, then the system can evaluate bootable states, but the complexity increases and accuracy decreases due to ever-changing representations
Solution Approach 1:
The patent replaces the mechanical text extraction and matching system with an artificial neural network-based image recognition system. Instead of extracting text from screenshots and matching against dynamic lists, the system uses a trained neural network to directly analyze visual patterns in boot screenshots, eliminating the need for text processing and complex matching logic while improving accuracy and reducing system complexity
Solution Approach 2:
The patent changes the evaluation parameter from text-based matching to visual pattern recognition. By training the neural network on diverse boot screenshots representing different operating systems and states, the system learns to identify integrity conditions through visual features rather than text content, making the evaluation robust against ever-changing representations across different OS versions
2Adaptability or versatility
If traditional text extraction methods are used, then the system can identify boot states, but it requires continuous updates to handle ever-changing representations
Solution Approach 1:
The patent performs preliminary training of the neural network on a comprehensive dataset of boot screenshots from multiple operating systems and versions before deployment. This pre-training enables the system to recognize diverse boot states without requiring continuous retraining when new OS versions are released, significantly reducing the time loss associated with model updates while maintaining high adaptability
Solution Approach 2:
The neural network is designed with universal recognition capabilities that work across different operating systems and versions. By training on diverse data during the preliminary phase, the single model serves multiple functions for evaluating various boot states without needing system-specific configurations or frequent retraining, achieving both adaptability and time efficiency
Data Source
AI summary
Described screenshot verification systems and methods for automatically verifying the integrity of a backup image or other process-of-interest using a screenshot verification system, as well as disaster recovery systems including said systems and performing said methods. In accordance with various aspects of the present disclosure, a virtual machine is booted and screenshots of the boot process are taken, which are used by a trained model, such as a convolutional neural network, to determine a boot state consistency. The systems and methods described deliver over 99% accuracy and do not involve regular expression analysis typical of conventional methods.