Real-name verification method for anchor identity in live broadcast

Abstract

The invention discloses a real-name verification method for an anchor identity in live broadcast. The method specifically comprises the following steps: (1) constructing a human face model and an eyebrow shape model of an anchor: separating human faces in video data to form human face model data of machine learning, and learning and further extracting the video data to make eyebrow shape model data; (2) when the third-party real-name verification system carries out verification, if the verification succeeds, ending the verification; after the verification fails, performing increased weight reasoning verification on the part above the eyebrows by using the video data, and if the verification succeeds, ending the verification; after the verification fails, verifying the eyebrow shape, and ifthe verification succeeds, ending the verification; entering the next step after the verification fails; and (3) establishing a face database of the anchor through manual confirmation, and correctinga machine learning model. The beneficial effects of the method are that the method greatly improves the recognition rate, reduces the cost of manual real-name check, and improves the broadcast starting efficiency of an anchor.

Description

technical field [0001] The invention relates to the technical field of video processing, in particular to a real-name verification method for the identity of an anchor in a live broadcast. Background technique [0002] Face recognition is a biometric technology for identity recognition based on facial feature information. By collecting video images and confirming whether there is a human face, it is a method for locating, collecting, identifying and confirming the identity of the human face. Since this technology is based on human facial features, the video image exists on the basis of the human face, find out the position, size and geometric information of each major facial organ of the human face, and form a collection of these information, and extract the corresponding data features from it , compared with the face database to identify the identity of the face. [0003] The streaming data in modern live broadcasts, the images of the anchors are all processed by beautifi...

AI Technical Summary

Problems solved by technology

[0003] The streaming data in modern live broadcasts, the images of the anchors are all processed by beautification, which can realize the functions of thin face, small face, big eyes, blush, and whitening. Jiemao performed make-up such as eye expansion. The wide application of these technologies and makeup made the face data formed by live broadcast collection and the existing face database be compared with special feature points. Deviations occurred, resulting in the failure of the real-name authentication of the anchor

[0004] Therefore, in view of the defect that the current existing technology is unable to compare the real-name of the face database collected by no beauty and other makeup techniques, and most of the real-name verification is a third-party program, it is necessary to perform face-based real-name verification. Research on technology to provide a solution to make up for the defects in the existing technology and avoid the situation that the anchor cannot start broadcasting

Images