Automated Video Categorization via Multi-Attribute Feature Ranking

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for ranking digital media assets are prone to inaccuracies due to issues like unmodeled objects, scenes, and content quality changes, and lack user-prioritized interest-based ranking.

Innovation Solution

A system computes feature attributes for digital media assets and applies a custom digital media value profile to create a ranked order, weighting attributes to prioritize user-interest, using techniques like SVM and CNN for accurate sorting.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If traditional video categorization methods are used, then videos can be organized by basic features, but user-prioritized interest-based ranking cannot be achieved

Engineering Contradiction:

Improveuser-prioritized interest-based rankingVSAvoidmulti-attribute feature computation system

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent segments video analysis into multiple independent attribute dimensions (visual features, audio features, metadata features) that can be computed separately and then combined through weighted aggregation to achieve user-prioritized ranking

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent creates a universal ranking framework that can handle multiple types of media assets (videos, images, audio) and supports customizable user profiles with different weighting schemes for various attribute types

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Measurement precision

If simple ranking parameters are used, then the system remains simple, but inaccuracies occur due to unmodeled objects and scenes

Engineering Contradiction:

Improveranking accuracyVSAvoidfeature computation and profiling system

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent transforms the ranking problem from simple parameter comparison to multi-dimensional feature space analysis, where each media asset is represented by a vector of attributes that can be precisely measured and compared according to user-defined weightings

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If comprehensive feature analysis is performed, then ranking accuracy improves, but processing time increases

Engineering Contradiction:

Improveranking accuracyVSAvoidprocessing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary feature extraction and stores computed attributes in a structured format, allowing rapid retrieval and re-ranking when user profiles change without re-processing the entire media library

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS10402436B2Automated video categorization, value determination and promotion/demotion via multi-attribute feature computation

Publication Date: 2019.09.03 PIXEL FORENSICS

AI summary

Techniques for automatically rank-ordering media assets based on custom-user profiles by computing features attributes from the media assets, and then applying a custom profile using its attribute weights and signs to create a final promotable value coefficient for each media asset. Then, using the value coefficient for each asset, a triage-able ranked order to the media assets can be created, with those assets the profile determines most promotable appearing first to the user.