Adaptive Video Ad Detection Using Watermark Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting televised video advertisements in real-time are inefficient due to high error rates, computational intensity, and inability to differentiate between ads and programs across multiple channels and languages, leading to synchronization issues in ad campaigns.
Innovation Solution
A computer-implemented method that adaptively switches between detection strategies based on the presence or absence of watermarked features, using keypoint extraction, binary descriptor creation, and vocabulary matching to identify advertisements in live media streams, with a focus on scale, rotation, and translation invariance, and progressive scoring for accurate ad detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If audio fingerprinting is used for ad detection, then ad detection capability is provided, but error rate increases and detection time increases
Solution Approach 1:
The patent extracts and removes watermarked features (logos, certificates, watermarks) from video frames to detect advertisement transitions. This extraction-based approach replaces the audio fingerprinting method, enabling faster detection without the error-prone matching processes that cause delays and inaccuracies in traditional audio-based systems.
Solution Approach 2:
The patent substitutes the audio-based detection mechanism with a visual-based mechanism using computer vision techniques. By replacing audio fingerprinting with image processing and watermarked feature detection, the system achieves both faster detection times and higher accuracy through direct visual analysis of advertisement content.
2Reliability
If audio fingerprinting is used for ad detection, then ad detection capability is provided, but false positives increase
Solution Approach 1:
The system extracts distinct watermarked features that are uniquely embedded in advertisement content. By focusing on these extracted visual markers rather than audio patterns, the system achieves more reliable detection with fewer false positives, as watermarked features provide explicit identification signals that are less prone to misinterpretation.
Solution Approach 2:
The patent detects changes in visual characteristics of video frames, including color and appearance variations that occur during advertisement transitions. These visual changes, particularly the presence or absence of watermarked features, provide clear differentiation between program content and advertisements, reducing false positive detections.
3Reliability
If supervised audio fingerprinting is used, then ad detection is performed, but computational efficiency decreases
Solution Approach 1:
The patent employs extraction of pre-defined watermarked features from video frames, which is computationally more efficient than full audio fingerprinting analysis. This extraction approach focuses processing on specific, predetermined visual markers rather than analyzing entire audio spectra, significantly improving computational efficiency while maintaining detection reliability.
Solution Approach 2:
The system performs preliminary processing by pre-defining and extracting specific watermarked features from video content before detailed analysis. This preliminary extraction step prepares the data in an optimized format that reduces subsequent computational requirements, enabling faster and more efficient ad detection compared to comprehensive audio fingerprinting methods.
4Measurement precision
If language-specific detection is used, then detection precision for specific languages is improved, but platform agnosticism is lost
Solution Approach 1:
The patent implements a universal detection system based on visual watermarked features that function across multiple platforms, languages, and regions. By using language-agnostic visual markers embedded in advertisement content, the system maintains platform agnosticism and versatility while achieving precise detection through the universal presence of these watermarked features regardless of the broadcast language.
Data Source
AI summary
The present disclosure provides a computer-implemented method and system for adaptively switching supervised detection strategy for watermarked and non-watermarked real time televised video advertisements. The televised video advertisements are present in a live stream of a media content of a broadcasted channel. The method includes selection of a set of frames per second from the media content. The method includes checking for one or more watermarked features in each selected frame of the media content. The method includes switching to a first detection strategy. The first detection strategy is associated with detection of a first ad in the live stream of the media content when the one or more watermarked feature are present in each checked frame in the selected set of frames.