Facial Expression Vectors for Temporal Emotional Change Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AI-based facial analysis technologies primarily focus on recognizing emotional states from facial expressions but fail to detect the temporal location of emotional changes, which are crucial for understanding the context and background of a person's feelings, especially in human-robot or human-computer interactions, and for individuals with autism spectrum disorder.
Innovation Solution
An apparatus and method for detecting emotional changes through facial expression analysis using a RGB camera, involving a preprocessing unit to extract and normalize face images, a facial expression analysis unit to generate a facial expression vector, and an emotional change analysis unit to identify the temporal location of sudden emotional changes by analyzing cumulative facial expression vectors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional AI-based facial analysis is used to recognize emotional states, then emotional state classification is achieved, but temporal location of emotional change cannot be detected
Solution Approach 1:
The patent segments the continuous emotional signal into discrete emotional change points by detecting significant transitions between emotional states. This segmentation approach allows the system to identify specific temporal locations where emotional changes occur, thereby recovering the temporal information that would otherwise be lost in continuous emotional state recognition.
Solution Approach 2:
The patent adds a temporal dimension to the traditional emotional state recognition by introducing time-based analysis. It detects emotional change points by analyzing the temporal evolution of emotional signals, transforming the problem from static emotional classification to dynamic temporal pattern recognition, thus preserving temporal location information.
2Device complexity
If facial expression analysis is performed without temporal change detection, then system complexity is reduced, but contextual understanding capability deteriorates
Solution Approach 1:
The patent extracts only the critical temporal change points from the continuous emotional signal rather than analyzing the entire signal continuously. By taking out only the significant emotional transition moments, the system maintains contextual understanding capability while avoiding the complexity of continuous temporal analysis throughout the entire signal duration.
3Loss of information
If emotional change detection is added to facial analysis, then responsiveness information is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary action by pre-defining emotional state categories and change detection thresholds before actual processing. This preparation allows the system to quickly compare incoming facial expressions against predefined criteria, enabling rapid detection of emotional changes without requiring complex real-time computations, thus minimizing processing time while maintaining responsiveness information.
Data Source
AI summary
Disclosed herein is an apparatus and method for detecting an emotional change through facial expression analysis. The apparatus for detecting an emotional change through facial expression analysis includes a memory having at least one program recorded thereon, and a processor configured to execute the program, wherein the program includes a camera image acquisition unit configured to acquire a moving image including at least one person, a preprocessing unit configured to extract a face image of a user from the moving image and preprocess the extracted face image, a facial expression analysis unit configured to extract a facial expression vector from the face image of the user and cumulatively store the facial expression vector, and an emotional change analysis unit configured to detect a temporal location of a sudden emotional change by analyzing an emotion signal extracted based on cumulatively stored facial expression vector values.