Ad Detection via Keyframe Binary Descriptors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting televised video advertisements are inefficient due to high error rates, computational inefficiency, and lack of language and platform agnosticism, leading to increased detection time and inaccurate differentiation between programs and ads.
Innovation Solution
A computer-implemented method for real-time detection of multi-lingual televised advertisements using keyframe selection, keypoint extraction, binary descriptor derivation, and spatial pyramid creation, with progressive scoring and clustering to identify ads across multiple channels and languages.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If supervised audio fingerprinting techniques are used for ad detection, then ad detection capability is provided, but detection time increases and error rate increases
Solution Approach 1:
The video stream is segmented into individual frames, and a subset of keyframes is selected for processing. This segmentation allows the system to process only representative frames rather than the entire continuous stream, reducing detection time while maintaining accuracy through strategic frame selection.
Solution Approach 2:
The system performs preliminary actions by pre-selecting keyframes and pre-processing them to extract visual features before actual ad detection begins. This preliminary extraction of visual descriptors from keyframes prepares the data in advance, enabling faster comparison and detection during runtime.
2Reliability
If audio fingerprints are used for ad detection, then ad identification is enabled, but false positives increase due to similar audio tones in non-ad content
Solution Approach 1:
The system merges multiple detection modalities by combining visual feature extraction from keyframes with audio analysis. This multi-modal approach cross-validates ad detection, reducing false positives that occur when relying solely on audio fingerprints by requiring concordance between visual and audio evidence.
Solution Approach 2:
Visual features extracted from keyframes serve as an intermediary validation layer between the audio fingerprint and the final ad detection decision. This intermediary visual analysis helps distinguish actual ads from non-ad content that may contain similar audio tones, reducing false positives.
3Reliability
If manual detection methods are used, then ad monitoring is performed, but efficiency is low and scalability is limited
Solution Approach 1:
The system replaces manual mechanical detection processes with automated computer vision and machine learning algorithms. Visual features are extracted automatically from keyframes using algorithms like SIFT or ORB, and ad detection is performed through automated comparison with a database of known ad templates, dramatically improving efficiency and scalability.
Solution Approach 2:
The system enables self-service detection by automatically selecting keyframes, extracting visual features, comparing them against the ad database, and generating detection results without human intervention. This automated pipeline processes multiple channels simultaneously, improving productivity while maintaining reliability.
4Measurement precision
If language-specific detection methods are used, then local ad detection is achieved, but language and platform agnosticism is lost
Solution Approach 1:
The system achieves universality by using visual feature extraction that is independent of language and platform. The visual descriptors extracted from keyframes (such as SIFT or ORB features) are language-agnostic and can detect ads across multiple languages and broadcasting platforms simultaneously, maintaining both precision and adaptability.
Data Source
AI summary
The present disclosure provides a computer-implemented method and system for hardware, channel, language and ad length agnostic detection of multi-lingual televised advertisements. The detection is performed across live streams of media content of one or more broadcasted channels. The method includes selection of a set of frames per second from a pre-defined set of frames. The method includes extraction of a pre-defined number of keypoints from each selected frame and derivation of a pre-defined number of binary descriptors from the extracted keypoints. The method includes creation of a special pyramid of the binary descriptors and accessing a second vocabulary of binary descriptors. The method includes comparison of each spatially identifiable binary descriptor from the first vocabulary with spatially identifiable binary descriptors in clusters of the second vocabulary. The method includes progressively scoring each selected frame and detection of the first ad in the live streams of the media content.