Dynamic Ad Insertion via Scene Metadata Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current advertising technologies fail to display relevant advertisements to users based on specific scenes within video content, as they primarily rely on metadata that describes a program as a whole rather than individual segments.
Innovation Solution
A method for dynamic advertising insertion that involves obtaining a video program with segments, extracting metadata for specific segments, selecting advertising content based on that metadata, and generating an output video stream with the selected advertisements displayed in connection with the corresponding segments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If advertising content is selected based on whole-program metadata, then the advertising system is simple to operate, but the advertisements are not relevant to specific scenes being viewed
Solution Approach 1:
The video program is divided into multiple segments (scenes), and metadata is extracted for each individual segment rather than treating the entire program as a single unit. This segmentation enables scene-specific advertisement selection, improving advertising relevance to what the user is currently viewing.
Solution Approach 2:
Metadata describing each video segment is extracted and stored in advance before advertisement selection occurs. This preliminary extraction of scene-specific metadata enables rapid, relevant advertisement matching when a segment is being viewed, without requiring complex real-time analysis during playback.
2Productivity
If advertisements are displayed based on whole-program metadata, then the system requires minimal processing, but user engagement and conversion rates are reduced
Solution Approach 1:
Metadata extraction is performed in advance for all video segments, creating a ready-to-use database of scene descriptors. This preliminary processing allows the system to quickly match advertisements to current scenes during playback without requiring intensive real-time computation, thus improving advertising effectiveness while managing processing resources efficiently.
Solution Approach 2:
Instead of analyzing the entire video program in real-time, the system uses pre-extracted metadata copies that describe each segment's content, objects, actions, and context. These metadata copies enable rapid advertisement selection without requiring the system to re-process the actual video data during playback.
3Measurement precision
If scene-specific metadata extraction is implemented, then advertising relevance is improved, but the complexity of metadata processing increases
Solution Approach 1:
The video program is divided into multiple segments (scenes), and metadata is extracted for each individual segment rather than treating the entire program as a single unit. This segmentation enables scene-specific advertisement selection, improving advertising relevance to what the user is currently viewing.
Solution Approach 2:
Metadata describing each video segment is extracted and stored in advance before advertisement selection occurs. This preliminary extraction of scene-specific metadata enables rapid, relevant advertisement matching when a segment is being viewed, without requiring complex real-time analysis during playback.
Data Source
AI summary
In one aspect, an exemplary method for dynamic advertising insertion includes obtaining a video program comprising a sequence of segments; extracting metadata describing at least a specific one of the segments of the video program; based at least in part on the extracted metadata describing the specific segment, selecting advertising content corresponding to the specific segment; and generating an output video stream comprising the sequence of segments, wherein the selected advertising content is displayed in connection with the specific segment.