Masked Face Recognition Using Synthetic Mask Training
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional face recognition methods fail when individuals wear masks, necessitating mask removal for identification, which is inconvenient and risky during pandemics.
Innovation Solution
A method using deep learning models, specifically YOLO5Face and ArcFace, trained with mask-generated images, enhances face detection and recognition accuracy while wearing masks, involving steps for mask generation, face detection, and feature extraction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional face recognition methods are used, then full face features are available for identification, but the method fails when masks are worn
Solution Approach 1:
The system performs preliminary actions by generating synthetic masked face images before actual recognition occurs. The mask generation algorithm creates training data with various mask types, positions, and lighting conditions, enabling the deep learning model to learn masked face features in advance. This preliminary preparation allows the system to maintain high recognition accuracy when masks are worn during actual operation.
Solution Approach 2:
The system changes parameters by transforming normal face images into masked face images through controlled parameter modifications. The mask generation algorithm adjusts parameters such as mask position, orientation, lighting angles, and mask types to create diverse training samples. This parameter variation enables the model to generalize across different masking scenarios while maintaining recognition reliability.
2Measurement precision
If masks are removed for identification, then face recognition accuracy is maintained, but user convenience and safety are compromised
Solution Approach 1:
The system extracts and focuses on the visible facial features that remain exposed when masks are worn. Instead of requiring full face visibility, the model learns to extract discriminative features from uncovered areas such as eyes, forehead, and nose bridge. This extraction approach maintains detection accuracy while allowing users to keep masks on, preserving both precision and convenience.
Solution Approach 2:
The system performs preliminary learning on synthetic masked images that simulate various mask-wearing scenarios. By pre-training with diverse mask configurations, the model learns to accurately detect and recognize faces under masking conditions without requiring users to remove masks during actual operation, thus maintaining both accuracy and ease of use.
3Reliability
If deep learning models are trained with generated mask images, then recognition accuracy while wearing masks is improved, but training data generation complexity increases
Solution Approach 1:
The system creates synthetic copies of normal face images by overlaying digital mask representations. Instead of manually collecting diverse masked face images, the algorithm generates realistic masked versions by copying and transforming existing unmasked images with applied mask layers. This copying approach produces large volumes of diverse training data while avoiding the complexity of manual data collection and annotation.
Solution Approach 2:
The system replaces manual mechanical processes (physical mask application, manual image annotation) with automated algorithmic generation. The mask generation algorithm automatically creates labeled training images by computationally overlaying masks on face images and generating corresponding annotation data. This substitution reduces training complexity by eliminating manual intervention while producing sufficient diverse training data.
Data Source
AI summary
The present invention provides a method of masked face recognition from images by artificial intelligence technology including four steps: step 1: generating the images of face wearing a mask; step 2: training a deep learning model for face detection while wearing a mask; step 3: training the deep learning model for face feature extraction while wearing a mask; step 4: building a full pipeline of masked face recognition from images using the trained models from step 2, step 3, and some post-processing algorithms. The method aims to improve the accuracy of identity verification in the context of wearing masks has become popular and compulsory in life.