Automated Media Tagging via Weighted Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for tracking and tagging media assets used in advertisements are inefficient, as they require manual coding or tagging with company information, and lack effective mechanisms for identifying and assigning tags to new or unknown media.
Innovation Solution
A system that extracts audio and image features from media assets using multiple recognition technologies, weights these features, and uses them as queries to search a database of pre-tagged media, thereby determining tags for new or unknown media.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If manual coding or tagging is used for media assets, then company information can be associated with media, but the process is inefficient and cannot effectively identify new or unknown media
Solution Approach 1:
The system enables media assets to self-identify and self-tag by automatically extracting features (audio, visual, textual) and matching them against a database of pre-tagged media. The media assets essentially tag themselves without human intervention, resolving the contradiction between automation extent and productivity.
Solution Approach 2:
The patent replaces manual mechanical tagging processes with automated computer-based feature extraction and pattern recognition systems. The mechanical manual coding process is substituted with automated algorithms that extract audio features, visual features, and textual features to automatically assign tags.
2Measurement precision
If multiple recognition technologies are used to extract features, then tagging accuracy improves, but system complexity increases
Solution Approach 1:
The system segments the complex tagging task into separate independent modules: audio feature extraction, visual feature extraction, and textual feature extraction. Each module uses specialized recognition technologies for its specific domain, improving overall accuracy while managing complexity through modular architecture.
Solution Approach 2:
The system employs a universal feature extraction framework that handles multiple types of media features (audio, visual, textual) through a common processing architecture. This multi-functional approach allows the system to accurately tag diverse media assets while avoiding the complexity of separate dedicated systems for each feature type.
3Productivity
If automated feature extraction is implemented, then productivity increases, but the ability to handle new or unknown media challenges increases
Solution Approach 1:
The system performs preliminary action by pre-processing and indexing a large database of media features and their corresponding tags before actual tagging operations. When new media arrives, the system quickly extracts features and matches them against the pre-prepared database, enabling high-speed tagging of new media without difficulty.
Solution Approach 2:
The system incorporates feedback mechanisms where extracted features from new media are compared with existing database patterns, and the system continuously learns from new data. This feedback loop enables the system to adapt to new media types and challenges while maintaining high productivity through automated processing.
Data Source
AI summary
Example methods, apparatus, systems and articles of manufacture are disclosed to determine tags for unknown media using multiple media features. Disclosed examples extract features from portions of the unknown media. Disclosed examples weight the features based at least partially on respective recognition technologies used to extract the features to determine corresponding weighted features. Disclosed examples search a database of pre-tagged media with a combination of the weighted features to generate a list of suggested tags for the unknown media, the list of suggested tags including relevancy scores for respective ones of the tags in the list. Disclosed examples assign a tag from the list of suggested tags to the unknown media based on a comparison of the relevancy score for the tag to a threshold.