Video Ad Insertion via Object Frame Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current advertisement insertion systems are ineffective as they insert irrelevant ads into video programs, annoying users and requiring time-consuming manual annotation, which limits the accuracy and relevance of ad placement.
Innovation Solution
A system and method that utilize an object database, advertisement database, and matching module to insert relevant ads into video programs by matching object information with image frames, with an advertising scheduling module that calculates scheduling scores based on bidding information and user location to optimize ad placement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If advertisements are inserted at the beginning or end of video programs, then advertisement placement is simple, but advertisement relevance to video content is low and user annoyance increases
Solution Approach 1:
The video program is segmented into multiple image frames that are analyzed individually. The system divides the video content into discrete visual elements and matches advertisements to specific frames based on object recognition, rather than placing ads uniformly at beginning or end. This segmentation enables precise, context-relevant ad placement throughout the video.
Solution Approach 2:
The system performs preliminary analysis of video frames to identify objects, scenes, and contexts before ad insertion. By pre-processing the video content to extract meaningful visual information and pre-matching potential advertisements to relevant frames, the system ensures ads are inserted at optimal moments without disrupting the viewing experience.
2Reliability
If manual annotation is used for textual matching between ads and web pages, then advertisement relevance can be achieved, but the process is time-consuming and text information is scarce and ambiguous
Solution Approach 1:
The system replaces manual textual annotation with automated computer vision technology. Instead of human annotators manually tagging text in video frames, the system uses object recognition algorithms to automatically identify and classify visual elements, extract meaningful information from images, and match advertisements based on visual content analysis. This substitution eliminates time-consuming manual processes while improving accuracy.
Solution Approach 2:
The system creates visual representations and feature extracts from video frames that serve as substitutes for manual text annotations. By copying and analyzing key visual features, object characteristics, and scene attributes directly from image data, the system generates sufficient matching information without requiring manual text extraction or annotation.
3Reliability
If the whole video program is subject to manual annotation, then comprehensive ad matching is possible, but information from individual image frames cannot be accurately grasped and processing efficiency is low
Solution Approach 1:
The system segments the video program into individual image frames and processes each frame independently through automated object recognition. This frame-by-frame analysis enables comprehensive ad matching across the entire video while maintaining high processing efficiency, as parallel processing can be applied to multiple frames simultaneously without the bottlenecks of manual annotation.
Solution Approach 2:
The system performs self-service automated analysis of video frames without human intervention. The computer vision algorithms automatically identify objects, extract features, and generate matching scores for ad insertion decisions, eliminating the need for manual annotation while accurately grasping information from individual image frames and maintaining high processing throughput.
Data Source
AI summary
The invention provides a system and method for inserting advertising in a video program, wherein appropriate advertisements are displayed at scheduled times in the process of a video program, characterized by matching image frames of a video program with object information and providing matching scores when an object information corresponding to an image frame; and inserting an advertisement corresponding to the object information having the matching score into the corresponding image frame in the video program.