Machine Learning Video Quality Estimation as an OS or Cloud Service

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video playback applications provide inconsistent, unreliable, and subjective feedback on video quality, which is not timely or detailed, limiting the ability of video streaming and conferencing services to adapt video encoding for improved delivery.

Innovation Solution

Implementing a machine learning model-based video quality estimation as an operating system or cloud service to objectively and consistently assess video quality, providing fine-grained feedback without viewer input, and adjusting encoding settings accordingly.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If video quality feedback is collected through surveys or feedback prompts, then some feedback is obtained, but the feedback is inconsistent, unreliable, and intrusive

Engineering Contradiction:

Improvefeedback reliabilityVSAvoiduser interaction burden

Core Design Contradiction:

ReliabilityVSEase of operation

Solution Approach 1:

The system uses the video playback application's own decoded video data to automatically estimate quality metrics without requiring separate user feedback mechanisms. The application serves itself by extracting quality information from the video stream itself rather than relying on user responses.

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces mechanical user interaction (surveys, prompts) with an automated computational system that analyzes video data mathematically. Quality estimation is performed through algorithms that process video frames, compression artifacts, and encoded data parameters automatically.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If video quality feedback is collected after playback completes, then feedback is available, but it is not timely and is only high-level

Engineering Contradiction:

Improvequality feedback detailVSAvoidfeedback timing

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The system performs quality estimation continuously during video playback rather than waiting for completion. By analyzing video data in real-time as it is decoded, the system provides timely feedback about compression artifacts, blocking, and other quality metrics throughout the playback process.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent breaks down overall video quality into multiple constituent quality metrics, including compression artifact analysis, region-specific quality scores, and temporal quality variations. This segmentation allows detailed feedback about specific portions of the video stream at any given time.

Inventive Principle:

Principle #1Segmentation

3Adaptability or versatility

If different video playback applications provide quality feedback, then feedback covers multiple scenarios, but the feedback is subjective and varies widely between viewers

Engineering Contradiction:

Improvefeedback coverageVSAvoidfeedback consistency

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The system changes the parameters used for quality measurement from subjective user ratings to objective technical metrics. By measuring compression artifacts, blocking, and video quality metrics through standardized algorithms, the system achieves consistent results across different applications and viewers regardless of their preferences.

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The quality estimation framework is designed to work universally across different video playback applications, streaming services, and device types. The same automated system can analyze video quality in video streaming, conferencing, and other scenarios without requiring application-specific feedback mechanisms.

Inventive Principle:

Principle #6Universality (Multi-functionality)

4Extent of automation

If video quality estimation is performed using engagement time as a proxy, then feedback is obtained without viewer input, but it is not timely or detailed and only loosely correlates with actual quality

Engineering Contradiction:

Improvefeedback collection automationVSAvoidquality measurement accuracy

Core Design Contradiction:

Extent of automationVSMeasurement precision

Solution Approach 1:

The patent replaces engagement time as a proxy metric with direct analysis of video data parameters. The system actually measures compression artifacts, blocking, and quality metrics from the video stream itself rather than inferring quality from viewing duration or engagement patterns.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system introduces an intermediary layer of automated quality analysis that sits between the video encoding process and the feedback collection system. This intermediary continuously monitors video quality metrics and provides detailed, timely information about actual video quality independent of viewer engagement.

Inventive Principle:

Principle #24Intermediary (Mediator)

Data Source

PatentUS20250287015A1Video quality estimation with a machine learning model as an operating system service or cloud service

Publication Date: 2025.09.11 MICROSOFT TECHNOLOGY LICENSING LLC

AI summary

With video quality estimation provided as an operating system service or cloud service, estimates of video quality can be collected unobtrusively and without feedback from video playback applications or viewers. For example, for a portion of reconstructed video content, an operating system service of a client computer system receives video data, estimates video quality of the portion of reconstructed video content using the video data, and sends results of the video quality estimation to an application executing on the client computer system. Or, as another example, for a portion of reconstructed video content, a cloud service of a server computer system receives video data, estimates video quality of the portion of reconstructed video content using the video data, generates encoder control values based at least in part on analysis of results of the video quality estimation, and sends the encoder control values to a streaming or conferencing service.