Emotion Recognition Using Discriminative Action Unit Relation Matrix
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current AU detection methods are prone to errors in tracking facial expressions and extracting features, leading to inaccurate emotion recognition when erroneous action units are mapped to target emotions.
Innovation Solution
A method that forms a relation matrix based on the discriminative power of Action Units (AUs) with respect to emotions, using a longest common subsequence technique to match input AU strings with template AU strings, allowing for insertion and deletion of AUs while suppressing substitution, to accurately recognize emotions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current AU detection methods are used to track facial expressions and extract features, then the system can process facial expressions, but the detection accuracy deteriorates due to errors in tracking and feature extraction
Solution Approach 1:
The system pre-processes the detected AUs by forming a relation matrix based on discriminative power before mapping to emotions. This preliminary processing step filters and weights AUs to reduce the impact of detection errors on final emotion recognition accuracy.
2Extent of automation
If formulated mapping rules are used to map AUs to target emotions, then the system can recognize emotions, but the system becomes sensitive to noise in the input AUs
Solution Approach 1:
The system changes the parameter of mapping by using a relation matrix with discriminative power weights instead of simple formulated rules. This allows the system to automatically handle noise by weighting reliable AUs higher and less reliable AUs lower in the emotion recognition process.
3Ease of operation
If the two phase method is used for recognizing facial expressions, then the system can map AUs to emotions, but it fails to compensate for weaknesses in AU detection techniques
Solution Approach 1:
The system introduces an intermediary relation matrix between AU detection and emotion mapping. This intermediary structure compensates for detection weaknesses by filtering and weighting AUs based on their discriminative power for different emotions, improving overall measurement precision while maintaining ease of operation.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An apparatus and method are provided for recognizing an emotion of an individual based on Action Units. The method includes receiving an input AU string including one or more AUs that represents a facial expression of an individual from an AU detector; matching the input AU string with each of a plurality of AU strings, wherein each of the plurality of AU strings includes a set of highly discriminative AUs, each representing an emotion; identifying an AU string from the plurality of AU strings that best matches the input AU string; and outputting an emotion label corresponding to the best matching AU string that indicates the emotion of the individual.