Is It Feasible To Make Face Recognition With A Mask?

Countries around the world have taken precautions since the outbreak of epidemic. Wearing surgical mask has become the most effective way to cut down the spread of the virus. However, citizens all wear masks will undoubtedly increase difficulty in face recognition scenarios such as face attendance, face access control where require face recognition. As mask will cover a large area of face, it is hard to accurately detect the face position and locate the key points of facial features. Thus is it feasible to make face recognition with a mask?

In fact, Baidu had proposed mask face recognition technology in 2018 and published related papers at the international top computer vision conference ECCV 2018 at the same year. The paper described that, Baidu face detection deep learning algorithm PyramidBox can detect a host of faces in public places and quickly identify people whether wear a mask. With the paper, Baidu obtained three awards in the WIDER FACE series' Easy ' 'Medium' and 'Hard' evaluation, which presented the world's most authoritative face detection credentials.

Based on the theoretical and driven by current epidemic outbreak situation, face depth recognition technology have been further applied in actual scenes. Telpo face recognition temperature measurement terminal is a good example.

 

It is reported that Telpo able to synchronously predict the face frame and landmark location through adjusting face detection deep learning algorithms, infrared thermal imaging temperature measurement algorithm and upgrading mask detection module. In virtue of converting face detection and alignment into standard key point estimation, Telpo machine can successfully make face recognition and accurately detect forehead temperature even people wear a mask.

Take Telpo face recognition temperature measurement terminal as an example. Employees wearing masks only need to stand in front of the terminal, and the terminal camera will accurately extract collected images’ feature data and compare it with the feature data in system photos to quickly and easily identify employees. Voice prompt is given to employees who do not wear a mask and the alarm mode is activated for abnormal temperature situation. The terminal background will mark and save the data in real time, reminding the epidemic prevention personnel synchronously and realize 24 hours supervision.

 

Telpo face recognition temperature measurement terminal has proved that it is feasible to make face recognition with a mask. It is suitable for customs, governments, enterprises, schools, communities, airports, subway stations, railway stations and other crowded public places. It solves the problem of taking off masks in public places by manual screening, reducing the risk of front-line workers infection and improves the screening efficiency.