Facial Key Point Evaluation Using Average Model Normalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for evaluating facial key point positioning results are time-consuming, labor-intensive, and lack efficiency, relying on manual intervention and subjective judgment, which reduces reliability and cannot quantify results effectively.
Innovation Solution
An evaluation method and device that acquire facial images and positioning result coordinates, perform normalization using an average facial model, extract facial feature values, and calculate evaluation results based on these values and a weight vector, eliminating the need for manual intervention and enabling quantitative evaluation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual marking of key point coordinates is performed, then evaluation accuracy can be achieved, but time consumption and labor costs increase significantly
Solution Approach 1:
The patent uses an average facial model as a reference template to automatically evaluate positioning results, replacing manual marking. The system copies the evaluation process from the standardized average model to individual facial images, enabling automated assessment without time-consuming manual coordinate marking while maintaining consistent evaluation criteria
Solution Approach 2:
The patent transforms the evaluation from manual coordinate-based assessment to automated feature-value-based assessment. By extracting facial features and comparing them with the average model parameters, the system changes the evaluation parameters from discrete coordinate points to continuous feature values, enabling efficient automated processing
2Productivity
If manual subjective evaluation is performed, then evaluation can be completed quickly, but the results cannot be quantified and reliability decreases
Solution Approach 1:
The patent implements an automated feedback mechanism where positioning results are objectively compared against the average facial model. The system calculates quantitative evaluation values based on feature differences, providing reliable and reproducible feedback without subjective human judgment, thus maintaining both speed and reliability
Solution Approach 2:
The patent replaces the mechanical process of manual subjective evaluation with an automated computational system. By using algorithmic processing to compare facial features with the average model, the system eliminates human subjectivity while maintaining high evaluation speed, achieving both quantitative results and reliability
3Measurement precision
If multiple positioning results are compared manually, then the most accurate result can be selected, but comparison difficulty and time consumption increase with large numbers of results
Solution Approach 1:
The patent transforms multiple positioning results from coordinate data into standardized evaluation values by comparing each result against the average facial model. This parameter transformation converts complex coordinate comparisons into simple value rankings, enabling automated selection of the most accurate result regardless of the number of candidates
Data Source
AI summary
Disclosed are an evaluation method and an evaluation device for a facial key point positioning result. In some embodiments, the evaluation method includes: acquiring a facial image and one or more positioning result coordinates of a key point of the facial image; performing a normalization process on the positioning result coordinate and an average facial model to obtain a normalized facial image; and extracting a facial feature value of the normalized facial image and calculating an evaluation result based on the facial feature value and a weight vector.