DiDi Open-Sourced the Industry’s First Mask-Wearing AI Detection Technology with an Accuracy Rate of 98%

DiDi Open-Sourced the Industry’s First Mask-Wearing AI Detection Technology with an Accuracy Rate of 98%

• DiDi’s AI detection technology has an accuracy rate of 98% in random field tests with in-vehicle cameras under various lighting environment and image composition conditions.
• Mask verification tool was first deployed in China in January and was later introduced to LatAm and APAC.
• DiDi has provided over 2 million trips serving frontline medical and relief workers across China and internationally as of June 2020.

(Beijing, July 2, 2020) – Didi Chuxing (“DiDi”), the world’s leading mobility and convenience platform, in its continued effort to better protect people from public health risks, released the industry’s first open-source AI model to verify whether individuals are wearing face masks.

Based on DiDi’s Depth First Search (“DFS”) face detection algorithm, the model achieves an accuracy rate of 98% in random field tests with in-vehicle cameras under various lighting environment and image composition conditions. The open-source AI tool will allow developers to use the detection system to analyze images, including mobile phone photos and in-vehicle camera images under various travel situations.

The detection technology was first deployed in China in January and expanded in June, under DiDi’s Health-Guard program, to LatAm and APAC. DiDi’s driver partners are required to take a photo for AI analysis and follow daily vehicle disinfection routines and report their body temperatures through the app. Driver partners who fail to follow these safety requirements may not be permitted to provide rides to users through DiDi’s platform.

In addition to face mask detection, DiDi has also deployed a series of protection and relief initiatives through its network, including a global driver and courier relief fund, special fleets for medical and community workers, vehicle sanitation stations, and in-vehicle protective sheets. DiDi has provided over 2 million trips serving frontline medical and relief workers across China and internationally as of June 2020.

“The industry-leading AI mask-recognition technology is designed to better protect users from public health risks,” said Dr. Ye Jieping, Head of DiDi AI Labs. “We are already talking to other ride-hailing companies and public interest organizations which need to ensure health management for large venues to provide support with this technology”.