TV Ad Detection via Subtitle Repetition and Object Association
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Television advertisers face challenges in quantifying the effectiveness of their advertisements and providing dynamic supplemental content, as existing systems lack the ability to accurately identify and enhance advertised products or services, especially when audio tracks do not include sufficient information, and viewers struggle to remember advertised items without additional support.
Innovation Solution
A system that analyzes television broadcast data, including caption and video frames, to identify repeated advertising content, detect advertising objects through image processing, and associate them with caption data, enabling the delivery of supplemental advertising information in real-time across various devices.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If television advertisements are broadcast to a large audience, then advertising reach is improved, but the ability to connect a particular product or service sale with a particular customer or airing of an advertisement deteriorates
Solution Approach 1:
The patent introduces an intermediary system that captures subtitle data from television broadcasts and stores it in a database. This intermediary mechanism enables the system to later retrieve and match subtitle information with viewer data, thereby establishing a connection between advertisements and customers without requiring direct tracking during the broadcast. The subtitle database serves as the mediator that bridges the gap between mass broadcast and individual customer connection.
2Duration of action of moving object
If television advertisements are created prior to products receiving awards or favorable reviews, then advertising timing is improved, but the ability to supplement with additional content describing awards and reviews deteriorates
Solution Approach 1:
The patent implements a dynamic content supplementation system where the subtitle database stores historical advertisement subtitles, and the system dynamically retrieves relevant subtitles based on current product events such as awards or reviews. This dynamic retrieval mechanism allows the system to adapt and supplement existing advertisements with new content (such as award information) without requiring the original advertisement to be recreated. The system maintains timing flexibility while gaining adaptability through on-demand content retrieval from the database.
3Ease of operation
If viewers watch television advertisements without additional support, then viewing simplicity is improved, but the ability to remember advertised products or take action deteriorates
Solution Approach 1:
The patent implements a feedback mechanism where the system captures subtitle data from advertisements and makes it available to viewers through various interfaces. This feedback loop allows viewers to access product information, award details, and supplemental content related to advertisements they watched. The system provides feedback to viewers in the form of accessible information that helps them remember products and take action, while maintaining the simplicity of the television viewing experience itself.
Data Source
AI summary
Methods, systems, and apparatus, receive television broadcast data for a plurality of television channels including caption data that includes text that corresponds to audio content of programming portions and advertising portions, and frame data defining video frames; identify as repeated caption data caption data that has been received at least a threshold number of times and received over at least a threshold number of channels; identify, from the repeated caption data, portions of broadcast data that correspond to the repeated caption data; select video frames from the portions; provide the selected video frames to a video processing system that identifies objects in the video frames and receiving from the video processing system the objects; and for at least one object, associate the object with the repeated caption data that corresponds to the video frame from which the object was identified.