AI Scene Gaze Estimation via Neural Network Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current technologies for evaluating viewer engagement with video content fail to accurately estimate interest in specific scenes, leading to inaccurate content recommendations and product promotions, as they rely on metadata associated with the entire content rather than individual scenes.
Innovation Solution
An artificial intelligence information processing device and method that uses neural networks to estimate the degree of gaze and scene information by correlating sensor data with video content, allowing for the identification and characterization of specific scenes a user is interested in, thereby providing more precise metadata for those scenes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If metadata associated with entire content is used for evaluation, then content recommendation can be provided, but accuracy of scene-specific interest estimation deteriorates
Solution Approach 1:
The patent segments the content into individual scenes and processes each scene separately. The gaze degree estimation is performed for each scene independently, and scene information is extracted and processed on a per-scene basis. This segmentation enables accurate scene-specific interest estimation while managing complexity through modular processing of discrete scene units rather than analyzing entire content at once.
2Measurement precision
If sensor information is processed to estimate gaze degree, then user engagement measurement is achieved, but processing time and computational load increase
Solution Approach 1:
The patent performs preliminary actions by pre-processing sensor information and pre-segmenting content into scenes before the actual gaze estimation. Scene information such as timestamps, duration, and content descriptors are extracted in advance. This preliminary preparation reduces the computational burden during real-time gaze degree estimation, thereby reducing processing time while maintaining measurement precision.
3Measurement precision
If neural network is used for scene information estimation, then estimation accuracy according to artificial intelligence is improved, but device complexity and computational resources required increase
Solution Approach 1:
The neural network processing is segmented into distinct stages: scene segmentation, feature extraction, and classification. Each stage processes specific aspects of the scene independently. This modular neural network architecture improves estimation accuracy for each scene while managing overall system complexity through organized, stage-wise processing rather than monolithic complex processing.
Data Source
AI summary
An artificial intelligence information processing device that generates information about a scene according to artificial intelligence is provided. The artificial intelligence information processing device includes a gaze degree estimation unit configured to estimate a degree of gaze of a user who is watching content according to artificial intelligence on the basis of sensor information, an acquisition unit configured to acquire a video of a scene at which the user gazes in the content and information about the content on the basis of an estimation result of the gaze degree estimation unit, and a scene information estimation unit configured to estimate information about the scene at which the user gazes according to artificial intelligence on the basis of the video of the scene at which the user gazes and the information about the content.