Facial Expression Recognition via Segmented Mask Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current systems face challenges in accurately and efficiently detecting facial expressions in real-time due to the complexity and computational costs of existing neural networks, making it difficult to implement effectively in applications like marketing and security.
Innovation Solution
A system comprising a facial mask generator to detect facial points and a learning network with a Squeezenet architecture and dual-stage structure, utilizing 3D geometrical and spatial facial information, and a recurrent input for classification, which generates a probability of facial expressions using a softmax function.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex neural networks are used for facial expression detection, then recognition accuracy is improved, but computational cost and system complexity increase
Solution Approach 1:
The system segments the facial expression detection task into two distinct stages: a detection stage that identifies facial points and generates masks, and a recognition stage that classifies expressions. This segmentation allows each stage to be optimized independently, reducing overall system complexity while maintaining accuracy.
Solution Approach 2:
The system extracts and utilizes 3D geometrical and spatial facial information as a separate feature input, rather than relying solely on raw image data. This extraction of key geometric features simplifies the processing required in the recognition stage while improving recognition accuracy.
2Measurement precision
If complex neural networks are used for facial expression detection, then recognition accuracy is improved, but processing speed decreases
Solution Approach 1:
The system performs preliminary detection of facial points and generation of facial masks before the recognition stage. This preliminary action prepares structured data that accelerates the subsequent recognition process, enabling real-time processing while maintaining high accuracy.
Solution Approach 2:
By dividing the task into detection and recognition stages, the system can optimize each for its specific function. The detection stage prepares data structures that make the recognition stage faster, achieving both accuracy and speed.
3Measurement precision
If detailed facial analysis is performed, then recognition accuracy is improved, but computational cost increases
Solution Approach 1:
The system extracts specifically 3D geometrical and spatial facial information, focusing computational resources on the most discriminative features for expression recognition. This selective extraction reduces overall computational energy consumption while maintaining or improving accuracy.
Solution Approach 2:
The system applies different processing quality to different aspects of facial analysis: detailed 3D geometric analysis for key facial points, and standard processing for other regions. This local quality approach optimizes energy usage by applying high computational effort only where most needed.
Data Source
AI summary
A system and method for determining a facial expression including a facial mask generator arranged to detect a plurality of facial points from one or more facial images to generate one or more associated facial point masks, arranged to represent the facial features of an associated face of the one or more facial images; and a learning network arranged to receive and input of the one or more facial images and the one or more associated facial point masks to obtain a classification result for each of the one or more facial images, wherein the classification result is one of a plurality of predefined facial expressions.