Facial Expression Recognition via Hierarchical Grouping
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current face expression recognition technologies face challenges in accurately recognizing human emotions due to the high variability of expressive behavior, which is influenced by context and domain, leading to inaccurate categorical representations and interference between expression categories.
Innovation Solution
A method and device that utilize cross entropy to calculate similarities between face expression categories, classify them into groups, and train convolutional neural network (CNN) models to improve recognition accuracy by merging similar categories and training separate models for each group, allowing for more precise classification of face images into expression categories.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current expression recognizers use simple pattern recognition techniques with categorical labels, then the device complexity is low, but the recognition accuracy deteriorates due to high variability of human expressive behavior
Solution Approach 1:
The patent segments the expression recognition task into two stages: first grouping similar expression categories together using similarity metrics, then training separate recognition models for each group. This segmentation allows the system to handle high variability within groups while maintaining lower complexity compared to a single comprehensive model.
Solution Approach 2:
The patent introduces dynamic adaptability by training multiple specialized models for different expression groups rather than using a single static model. Each model is dynamically optimized for its specific group, allowing the system to adapt to the high variability of human expressions while managing complexity through specialization.
2Reliability
If categorical representations are used for expressions, then the device complexity is low, but the reliability deteriorates due to interference between expression categories
Solution Approach 1:
The patent segments expression categories into distinct groups based on similarity metrics, reducing interference between categories. By processing expressions in grouped categories rather than as isolated labels, the system improves reliability while managing complexity through structured organization.
Solution Approach 2:
The patent introduces similarity metrics and expression groups as intermediary structures between raw expression data and final recognition results. These intermediaries help manage category interference by providing a structured framework that reduces direct conflicts between similar expression categories.
3Measurement precision
If separate models are trained for each expression group, then the recognition accuracy improves, but the device complexity increases due to multiple models
Solution Approach 1:
The patent segments the recognition task into hierarchical levels: expression grouping and then group-specific recognition. This segmentation improves classification accuracy by allowing specialized models for each group while managing overall complexity through the structured hierarchical approach.
Solution Approach 2:
The patent changes the organizational parameter from individual expression categories to expression groups based on similarity metrics. This parameter change allows multiple models to be managed more efficiently by grouping similar expressions together, improving accuracy while controlling complexity through meaningful organization.
Data Source
AI summary
A method for recognizing face expressions is provided. The method includes: recognizing expression categories of expressions in a plurality of face images and obtaining recognition results between each expression category and another expression category; obtaining similarities between each expression category and another expression category according to the recognition results; classifying the expression categories into a plurality of expression groups according to the similarities; training a first recognition model to classify the expressions in the face images into the expression groups; and training a second recognition model for each of the expression groups to classify the face images in each of the expression group into one of the expression categories.