Non-Fingerprint Content Recognition via Element Detection Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing automatic content recognition (ACR) technologies rely on fingerprinting and watermarking, which are costly, resource-intensive, and face challenges with high-definition content processing, particularly in edge-cloud systems, due to the need for large reference libraries and processing delays.
Innovation Solution
The implementation of a non-fingerprint-based ACR system that uses computer-implemented sequences of rules and logic to detect and identify content elements by sampling and iteratively matching using element detection models and machine learning, eliminating the need for content-based reference libraries and reducing resource consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If fingerprinting and watermarking techniques are used for content recognition, then content identification accuracy is improved, but system complexity and resource consumption increase
Solution Approach 1:
The patent extracts only the necessary content elements (faces, people, vehicles, buildings, plants, animals, cities, geographic features, articles of clothing, signs, textual or numeric information, slogans, logos, symbols, locations, words, jingles, brands, trade names, trademarks, landmarks, visual works, audio works, audiovisual works) from the content using element detection models, rather than processing the entire content fingerprint. This extraction approach maintains identification accuracy while reducing system complexity.
Solution Approach 2:
The patent segments the content recognition process into distinct stages: sampling source content, detecting specific content elements using element detection models, generating bounding boxes, and identifying elements through machine learning. This segmentation allows each component to be optimized independently, reducing overall system complexity while maintaining accuracy.
2Measurement precision
If large reference libraries are used for content recognition, then identification accuracy is improved, but network resource consumption and processing time increase
Solution Approach 1:
The patent applies partial action by using element detection models to identify only specific content elements within the content rather than comparing the entire content against large reference libraries. This partial processing approach maintains identification accuracy while significantly reducing processing time and network resource consumption.
Solution Approach 2:
The patent performs preliminary detection of content elements using element detection models before identification. This preliminary action filters and prepares the data in advance, allowing faster matching and identification processes without requiring large reference libraries to be loaded and processed in real-time.
3Reliability
If fingerprinting methods are used for content recognition, then content identification is achieved, but overhead costs and resource consumption increase
Solution Approach 1:
The patent replaces the traditional mechanical fingerprinting system with an element detection and identification system using machine learning models. This substitution eliminates the need for creating and storing fingerprints, reducing computational overhead and resource consumption while maintaining reliable content recognition capability.
Solution Approach 2:
The patent uses lightweight element detection models that process content elements on-demand without requiring expensive, long-lived fingerprint databases. Each content element is detected and identified independently, allowing the system to use simpler, more resource-efficient processing methods while maintaining recognition reliability.
Data Source
AI summary
A method for content recognition. The method may include sampling a source content for performing content recognition; detecting content elements from the sampled source content; and identifying the detected content elements, wherein detecting the content elements from the sampled source content comprises: detecting the content elements using an element detection model; and generating bounding boxes over the detected content elements. In one exemplary embodiment, each bounding box corresponds to a detected content element, and each detected content element is a detected face, slogan, logo, symbol, location, word, jingle, brand, trade name, trademark, landmark, building, visual work, audio work, audiovisual work, and/or any other product or item of unique interest, etc.