slide 1


Function details

Face detection

It recognizes whether there is a human face in the image and gives a face area, and returns the detected rectangle coordinates of the face after face detection. It can support simultaneous detection of multiple faces, support up and down 60 degree rotation face detection, and support left and right maximum 30 degree side face detection. At the same time, the facial features of the face are extracted in milliseconds, and 512 key points of the face are identified.

Face matching

Compare the similarity of two faces and give a similarity score to determine whether they are the same person. At the same time,return the rectangular frame coordinates of the two faces, the confidence of comparison, and the confidence threshold of different error rate.

 

   

Face searching 1:N

The face image captured by the face capture IPC is compared with the face database to realize face recognition of 1:N. Applicable to scenes without card verification such as face attendance, VIP face recognition etc.Provide sdk calling mode, support no interactive detection, support multi-platform, CPU, GPU computing mode, flexible deployment, tens of millions of face sample collection, CNN network model continuous training and tuning, and efficiency and accuracy of identity comparison network.

Face recognition algorithm live demo

The user logs in to the webpage or pays attention to the WeChat public account to experience the face recognition algorithm, and supports 1:1 face comparison and million-level face database millisecond-level fast searching.