Face Recognition has set many records in international competitions. It achieved an 99.80% accuracy in LFW 2017 and an 83.29% recognition rate in the MegaFace challenge.
Face Recognition has been verified by a large number of users and complex use cases. It is proven to be highly reliable and robust with service availability up to 99.9%.
Based on the 3rd generation Tencent YouTu Grandmother Model, Face Recognition has integrated multiple training methods to optimize the model, including metric learning, transfer learning, multi-task learning, etc. It can custom fine-tuning or distilling models to meet requirements for performance and latency in different scenarios.
Face Recognition provides simple APIs and has diverse use cases such as access control, security surveillance, VIP identification, sign-in, payment and login.
Face Recognition features high concurrence, high throughput, and low latency. It can search and process millions of faces in hundreds of milliseconds, meeting you needs in real time.