Baidu Releases AI to Identify People Who Don't Wear Masks Amid Coronavirus

The open-source tool is able to identify people who are not wearing masks with 96.5% accuracy.
Derya Ozdemir
The photo credit line may appear like this1, 2

China is battling with the deadly coronavirus as it gets more and more widely-spread. The current toll is at 1,874 deaths and 73,425 cases worldwide with China shouldering the most of the numbers.

Since the virus can spread through close contact with an infected person, the government has made it mandatory for its citizens to wear face masks when visiting public places such as shopping malls and public transport. However, with a population as dense as China’s, it is impossible to regulate if everyone is following the rules.

There’ve been cases where drones told people to wear their masks, Chinese children reported their parents to the police because they don’t wear their masks, and now, the country’s leading search engine Baidu has created an open-sourced an AI model to detect people not wearing face masks. 


The face-scanning model uses artificial intelligence to identify people in real-time who are not wearing masks or those who are wearing them incorrectly, Baidu says. Apparently, the AI was trained with 100,000 images to achieve a prediction accuracy of 96.5%, which is enough for the daily inspections

This is great news since inspecting people who aren’t wearing masks in crowds is done manually, and thereby, extremely difficult. Detecting people who are risking the lives of others is especially important now with the end of the Lunar New Year holiday. Many workplaces have resumed working and the risk of catching the virus is still quite high.

Using such a tool can be useful with containing the virus, and prevent more people from becoming infected. However, more and more tech companies gaining access to people's faces remains an issue. 


Subscribe today

For full access to all features
and product updates.

%30 Save Quarterly




Subscribe Now
You can cancel anytime.
View Other Options

Already have an account? Log in

0 Comment
Already have an account? Log in