Facial Emotion Detection Using Neutral-Image Correction Values
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing emotion determination devices face challenges in precision due to individual differences in facial expressions and potential errors from image detection, leading to incorrect emotion outputs.
Innovation Solution
An emotion determination device that acquires a facial image, sets evaluation values for emotion classifications based on a neutral state, computes correction values using a neutral facial image, and applies these values to determine emotions, accounting for individual differences and divergences from a general model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If correction processing is performed to increase uniformity of evaluation values, then slight changes in facial expression are reflected in emotions, but image detection errors may cause incorrect emotions to be output
Solution Approach 1:
The system performs preliminary action by acquiring a neutral facial image before emotion detection and using it to generate correction values that compensate for individual differences. This preliminary correction value generation prevents detection errors from causing incorrect emotion outputs while maintaining sensitivity to actual emotional changes.
Solution Approach 2:
The system changes parameters by adjusting evaluation values based on correction values derived from neutral facial images. These correction values modify the threshold and baseline parameters for emotion detection, allowing the system to account for individual facial expression characteristics while maintaining reliable emotion determination.
2Device complexity
If general emotion detection models are used, then processing is simplified, but individual differences in facial expressions cause determination errors
Solution Approach 1:
The system applies local quality by generating individual-specific correction values based on each user's neutral facial image. This allows the general detection model to be locally adjusted for each user's facial characteristics, improving precision without requiring completely different detection models for each individual.
Solution Approach 2:
The neutral facial image serves as an intermediary that bridges the general emotion detection model and individual user characteristics. By extracting correction values from the neutral image, the system creates a mediator that adapts the general model to individual users without increasing overall system complexity.
Data Source
AI summary
An emotion determination device includes a memory and a processor connected to the memory. The processor is configured to acquire a facial image of a user, set an evaluation value for each of plural emotion classifications including a neutral state based on the acquired facial image, compute a correction value using an evaluation value set based on a neutral facial image acquired at a timing corresponding to a neutral state, and determine an emotion of the user by applying the computed correction value to the set evaluation values.