RISK IDENTIFICATION METHOD EMPLOYING FACIAL IMAGE, DEVICE, COMPUTER APPARATUS, AND STORAGE MEDIUM

摘要

A risk identification method employing a facial image, a device, a computer apparatus, and a storage medium. The method comprises: acquiring video data of a client answering a pre-determined question (S10); performing, by using a pre-trained blocking identification model, image identification on the video data, an identification result comprising images of a face blocked by a hand and images of the face not blocked by a hand (S20); acquiring a face position blocked by a hand in each image of the face blocked by a hand, and acquiring a nervousness level identification result on the basis of face positions blocked by a hand corresponding to all of the images of the face blocked by a hand (S30); inputting the images of the face not blocked by a hand into a pre-trained security determination model, and acquiring a risk level identification result (S40); and determining a risk warning identification result according to the nervousness level identification result and the risk level identification result, and if the risk warning identification result is greater than a first pre-determined threshold, then determining the video data as risk warning data (S50). The invention resolves the issue of inaccurate identification caused by a hand blocking a face during a video data analysis processes.