Neural Network Video Quality Assessment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video quality assessment methods for video-based applications in cloud environments are inflexible and fail to accurately reflect user experience, as they rely on mathematical metrics that do not align with human judgment, leading to suboptimal resource optimization and user experience.
Innovation Solution
A system and method that uses a neural network trained with augmented video data to assess video quality, generating both subjective and objective quality information, allowing for more accurate reflection of user experience and enabling optimized resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional mathematical metrics are used for video quality assessment, then the assessment process is simple and fast, but the accuracy of reflecting user experience deteriorates
Solution Approach 1:
The patent replaces traditional mathematical metrics (mechanical/systematic approach) with a neural network-based machine learning system that mimics human visual perception. The neural network is trained on labeled video data with subjective quality scores, enabling it to predict video quality in a way that aligns with human judgment rather than relying on rigid mathematical formulas.
Solution Approach 2:
The patent transforms the assessment approach by changing from fixed mathematical parameters to adaptive learned parameters through training. The neural network learns optimal feature representations and quality prediction parameters from training data, allowing the system to adapt to different video content and quality characteristics rather than using static mathematical metrics.
2Loss of energy
If cloud resources are optimized for cost saving, then operating costs are reduced, but video quality may deteriorate
Solution Approach 1:
The patent implements a feedback mechanism where the neural network continuously assesses video quality and provides quality metrics back to the system. This feedback enables dynamic adjustment of cloud resources - when quality is sufficient, resources can be reduced for cost savings; when quality degrades below acceptable levels, resources can be increased to maintain quality, thus resolving the trade-off between cost and quality.
Solution Approach 2:
The patent introduces dynamic resource allocation based on real-time quality assessment. Instead of static resource allocation, the system continuously monitors video quality using the neural network and adjusts cloud resource allocation dynamically, allowing optimization of both cost and quality by matching resources to actual needs rather than maintaining fixed high-level provisioning.
3Reliability
If more cloud computing resources are allocated to video-based applications, then video quality is improved, but operating costs increase
Solution Approach 1:
The patent applies partial action by allocating cloud resources proportionally to actual quality needs rather than providing excessive resources for all scenarios. The neural network identifies the minimum necessary resources to achieve acceptable quality levels, avoiding unnecessary resource allocation while maintaining sufficient video quality, thus reducing costs without sacrificing essential quality.
Data Source
AI summary
A system and method for assessing video quality of a video-based application trains a neural network using training data of video samples and assesses video of the video-based application using the neural network to generate the subjective video quality information of the video-based application. Data augmentation is performed on video data, which is labeled with at least one subjective quality level, to generate the training data of video samples.