Mask Wearing Detection Using Face Landmark Overlap
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing mask detection systems are time-consuming and lack accuracy in determining the standardization of mask wearing.
Innovation Solution
A method using a multi-task network model to quickly and accurately detect mask wearing by determining a target region based on comparison information and overlapping degree, supported by a customized comparison criteria.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a cascade of various stages (detection, classification, etc.) is used for mask detection, then detection accuracy can be maintained, but detection time increases significantly
Solution Approach 1:
The patent combines multiple detection tasks (face detection, mask detection, and standardization detection) into a single multi-task network model. This merging of previously separate cascade stages into one integrated model reduces the overall detection time while maintaining accuracy across all detection objectives.
Solution Approach 2:
The multi-task network model performs multiple functions simultaneously: detecting faces, detecting masks, and determining standardization compliance. This universal model replaces the need for separate specialized models for each detection stage, achieving both speed and accuracy through a single unified system.
2Adaptability or versatility
If traditional multi-stage detection systems are used, then comprehensive detection can be achieved, but the system complexity increases
Solution Approach 1:
The patent merges multiple detection functionalities into a single multi-task network model, reducing system complexity by eliminating the need for separate detection modules while maintaining comprehensive detection capabilities through the integrated model's multiple output tasks.
3Measurement precision
If face detection and mask detection are performed in separate stages, then each detection can be optimized independently, but overall detection time increases
Solution Approach 1:
The patent merges face detection and mask detection into a single multi-task network model that processes both tasks simultaneously. This integration maintains the accuracy benefits of specialized detection while achieving the efficiency gains of parallel processing within a unified model architecture.
Data Source
AI summary
The method includes: receiving an image to be detected, wherein the image to be detected includes an object to be detected; detecting the image to be detected based on a trained multi-task network model to obtain a region where the object wears a mask; comparing an area of the region where the object wears the mask to a predetermined threshold to determine a first detection result; in response to the object being determined as wearing the mask, determining comparison information of a mask area to be compared, and determining a target region in the image to be detected corresponding to the comparison information based on a face feature point template; and obtaining a second detection result based on an overlapping degree between the target region and the region where the object wears the mask.