Cloud Face Recognition Services

Authentication and Face-Recognition Services.

All identity management systems have methods for authenticating users, and indexing database entries. For authentication and sharing Animetrics is built on Openid and Oauth; for indexing Animetrics has built the Animetrics90 facial scoring engines for biometric indexing. The AnimetricsFR system is a patented 2D-3D face recognition system that is fully scalable via FastCGIs. The FR system utilizes proprietary, sophisticated human face analysis algorithms to identify faces in 2D images, to create a 3D avatar from one or more 2D images, and to create face templates. The face templates can be compared, or searched against, to compute similarity between an enrolled image or a watch list of enrolled images of people. The search against a watch list of images utilizes sophisticated indexing techniques to search against thousands of records.

Authentication Services and Application Layers

Shown in the accompanying figure are the REST service and PHP layers along with their interaction with the backend NoSQL database representing the FaceR Cloud.

FaceR Enrollment Services

Every facial search engine has an enrollment phase during which the 500:1 data reduction occurs generating the discriminating low-dimensional parametric representation which represents the original photograph. We call this parameter representation the Animetrics Face Signature. Shown in the accompanying figure is the enrollment service architecture.

The FaceR enrollment system provides a mechanism to register a human face with the FaceR face recognition system. The enrollment process involves submitting a set of 2D images to the FaceR server from a mobile device or a web browser (and can also be extended to any desktop application or any device that can image and send images via HTTP) along with a unique identification entity that identifies the person (e.g., a mobile device identifier). The FaceR server analyzes the images for any human face and associates the unique identification entity with the human face, creates a face template and stores the template in a fault-tolerant, redundant, secure, and, a scalable database.

FaceR Verification Services

All authentication and identity management system have an authentication or verification stage during which time the discriminating low-dimensional parametric representation which represents the original photograph, which we denote the Animetrics face signature, is scored via a similarity measure relative to an enrolled set of facial signatures. The FaceR verification system computes the similarity between a human face in a 2D image with a set of enrolled images. The verification process involves sending a 2D image from a device that is already registered (during the enrollment process) or a application that can send the unique identification entity to compare against, to the FaceR server. The FaceR server analyzes the 2D image for any human face and if a face was found, computes the similarity between the face in the image and the enrolled face.