Facial Recognition Using Dual Convolutional Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing facial recognition methods experience low accuracy when the face is partially covered, such as with sunglasses or a mouth-muffle, as they fail to effectively utilize partial facial features for recognition.
Innovation Solution
A method and apparatus that utilize pre-trained convolutional neural networks to extract complete and partial facial feature information, where the complete facial feature information and partial facial feature information are combined to improve recognition accuracy by representing the correlation between them, specifically using a first convolutional neural network to acquire complete and partial facial feature maps and a second convolutional neural network to generate a facial recognition result.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing facial recognition methods directly examine the entire facial area, then the recognition process is simple, but the accuracy is low when the face is partially covered
Solution Approach 1:
The patent divides facial recognition into two segments: complete face recognition and partial face recognition. The system extracts complete facial feature information and partial facial feature information separately, then combines them for final recognition. This segmentation allows the system to handle both fully visible faces and partially covered faces effectively, improving accuracy without excessive complexity.
Solution Approach 2:
The patent applies partial action by focusing on extracting and utilizing only the visible partial facial features when the face is covered, rather than requiring the entire face. The system extracts partial facial feature information from visible regions (eyes, nose, mouth) and combines it with complete face features when available, enabling recognition to proceed with incomplete information.
2Loss of information
If the system uses only complete facial features for recognition, then the processing is straightforward, but it fails to utilize partial facial features effectively
Solution Approach 1:
The patent merges complete facial feature information and partial facial feature information into a unified recognition framework. Both types of feature information are extracted by the first convolutional neural network and then fed into the second convolutional neural network for combined recognition, ensuring that no useful information is lost regardless of face coverage.
Solution Approach 2:
The patent creates a universal facial recognition system that can handle multiple scenarios: full face recognition, partial face recognition, and combinations thereof. The dual-branch feature extraction architecture allows the system to adaptively process different types of input (complete or partial faces) through the same recognition pipeline, making the system multi-functional.
Data Source
AI summary
Embodiments of the present disclosure disclose a method and apparatus for facial recognition. A specific embodiment of the method includes: acquiring a to-be-recognized image; inputting the to-be-recognized image into a pre-trained first convolutional neural network to obtain complete facial feature information and partial facial feature information, the first convolutional neural network being used to extract a complete facial feature and a partial facial feature; and inputting the complete facial feature information and the partial facial feature information into a pre-trained second convolutional neural network to obtain a facial recognition result, the second convolutional neural network being used to represent a correlation between the facial recognition result, and the complete facial feature information and the partial facial feature information. This embodiment improves the accuracy of the recognition result in a situation where a face is partially covered.