Deep Neural Network Brand Recognition in Live Video
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems fail to effectively quantify the exposure and screen time of brands in live broadcasted video content, particularly in sports events, due to inadequate training and testing, leading to poor memory retention and advertising value assessment for advertisers.
Innovation Solution
A method and system utilizing a trained machine learning model with deep neural networks, specifically convolutional neural networks (CNN) and region convolutional neural networks (R-CNN), to extract and classify brand features from video frames, tracking screen time and displaying information on brand recognition, categories, and frequency of displayed brands.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional brand detection systems are used in live broadcast video content, then brand identification can be performed, but the systems fail to accurately quantify screen time and exposure frequency of displayed brands
Solution Approach 1:
The system performs preliminary training of deep neural network models with extensive brand image data before deployment. The models are pre-trained to recognize diverse brand logos, products, and advertisements in various contexts, enabling accurate real-time detection during live broadcasts without requiring complex processing during the actual measurement phase.
Solution Approach 2:
The patent introduces an intermediary processing layer that bridges raw video frames and final brand exposure metrics. This layer includes multiple neural network components (CNNs, R-CNNs) that progressively extract features, identify brands, track their appearance across frames, and calculate screen time metrics, thereby mediating between video input and quantitative output.
2Measurement precision
If deep neural networks are implemented for brand recognition, then brand detection accuracy is improved, but system complexity and training requirements increase significantly
Solution Approach 1:
The brand recognition system is segmented into multiple specialized neural network components: initial CNN layers for feature extraction, R-CNN modules for region-based detection, and separate processing streams for different brand identification tasks. This segmentation allows each component to specialize in specific aspects of brand recognition, improving overall accuracy while enabling modular deployment and training.
Solution Approach 2:
Extensive preliminary training is performed offline using large datasets of branded content. The deep neural networks are pre-trained on diverse images containing various brands, products, and advertisements in different contexts. This preliminary action transfers learned features to the live broadcast analysis system, reducing the need for complex real-time processing and enabling accurate brand detection with established models.
3Loss of information
If comprehensive brand tracking is performed in live broadcasts, then advertising value assessment is enhanced, but processing time and computational resources increase
Solution Approach 1:
The system performs preliminary training and model preparation offline, establishing robust brand recognition capabilities before live broadcast processing. Pre-trained deep neural networks with extensive brand knowledge are deployed, enabling rapid real-time inference during broadcasts without requiring complex processing or extensive computation during the actual measurement phase.
Solution Approach 2:
The brand tracking system operates continuously throughout the live broadcast, maintaining constant monitoring of brand appearances across video frames. The neural network processes video streams in real-time, continuously identifying and tracking brands as they appear, ensuring complete capture of all brand exposures without interruption or sampling, thereby providing comprehensive advertising value assessment.
Data Source
AI summary
A method for brand recognition in video by implementing a brand recognition application coupled to a streaming media player, for identifying an observed set of brands streamed in a broadcast video; receiving, by the brand recognition application, a broadcast video stream of a series of images contained in consecutive frames about an object of interest; extracting a set of brand features from each of image received by applying a trained brand recognition model with neural networks in order to detect one or more features related to each displayed object of interest in each frame, wherein the object of interest is associated with a brand image contained video content displayed to a viewer; and displaying, by a graphic user interface, information from the brand recognition application comprising at least time detected of the brand image in the video content of the broadcast video.