Image Evaluation Using Face Position and Size Weighting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image evaluation methods fail to accurately reflect subjective viewer preferences when evaluating images based on facial features, as they rely solely on quantitative calculations rather than user-centric evaluation criteria.
Innovation Solution
An image evaluation device and method that statistically calculates evaluation values using pre-determined weighting functions based on face size, position, number, orientation, and distance from the image center, reflecting user preferences by summing temporary evaluation values for multiple faces in an image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image evaluation is performed using only quantitative calculations of facial features, then the evaluation process is simple and objective, but the results do not accurately reflect subjective viewer preferences
Solution Approach 1:
The patent applies feedback by collecting actual viewer evaluation results and using them to adjust and optimize the weighting functions. The system continuously improves its evaluation accuracy by incorporating real-world feedback data, where viewer preferences are used to refine the weighting coefficients in the evaluation model, creating a closed-loop system that learns from actual usage patterns
Solution Approach 2:
The patent changes parameters by introducing multiple weighting coefficients (w1, w2, w3, w4) that can be adjusted based on different evaluation scenarios and viewer preferences. These parameters allow the system to dynamically modify the importance of different facial features (face area ratio, eye open state, orientation, expression) to better match subjective evaluation criteria
2Measurement precision
If multiple facial features are evaluated with equal weight, then the evaluation is comprehensive, but it does not reflect the varying importance of different features to viewer preference
Solution Approach 1:
The patent applies local quality by assigning different weighting coefficients to different facial features based on their relative importance. Instead of treating all features equally, the system gives higher weights to features that viewers care about more (such as face area ratio and eye open state) and lower weights to less important features, creating a differentiated evaluation approach that reflects local variations in feature importance
Solution Approach 2:
The patent implements dynamics by making the weighting coefficients adjustable and adaptable. The weighting functions can be modified based on different image types, viewing contexts, and viewer preferences, allowing the evaluation system to dynamically adjust the importance of different facial features rather than using fixed static weights
3Measurement precision
If subjective viewer preferences are incorporated into image evaluation, then evaluation accuracy improves, but the evaluation process becomes more complex
Solution Approach 1:
The patent applies self-service by enabling the system to automatically learn and adapt to viewer preferences without requiring manual intervention for each evaluation. The weighting functions are automatically adjusted based on collected evaluation data, and the system self-optimizes its parameters to improve accuracy over time without requiring complex manual configuration or adjustment by users
Data Source
AI summary
More accurate image evaluation is achieved by using information about faces contained in images. An information acquiring unit acquires, from an image containing at least one face, information about the size and the position of the at least one face. An evaluation value calculating unit calculates an evaluation value representing a result of evaluation of the image based on the information about the size and position of the at least one face acquired by the information acquiring unit.