An edge scene living body data targeted screening method, device, equipment and medium

Abstract

The present application relates to the technical field of face recognition security, and relates to an edge scene live body data targeted screening method, device, equipment and medium, the method comprising: extracting corresponding sample scene feature vectors from image data of each training sample in a mixed training data set by using a scene feature extraction model, establishing a feature index library based on each sample scene feature vector, extracting target scene feature vectors from each target image data by using the scene feature extraction model, performing similarity calculation processing on each target scene feature vector and each sample scene feature vector in the feature index library to obtain a similarity score set, performing training processing and compression processing on a face live body detection model to be deployed based on a targeted training sample set to obtain a face live body detection model suitable for deployment on an edge device. The present application has the effect of improving the applicability of edge deployment under the condition that scene differences and resource constraints coexist.

Images