Heterogeneous video monitoring face recognition method based on edge calculation

The invention relates to a heterogeneous video monitoring face recognition method based on edge computing. The method comprises the steps of edge device face detection, face alignment, back-end service face image feature extraction and cloud database face feature comparison. A user-defined face recognition network model is deployed on an edge computing device, key points of a face are detected and obtained, and then face alignment operation is performed on the edge device and sent to a cloud server. And the cloud server obtains the aligned face image and the edge device id, carries out face feature extraction, and compares the obtained face features with the face features in the big data system to carry out face recognition and information retrieval. According to the method, the key points of the face are detected at the edge end and pushed, so that the network bandwidth pressure is reduced, the computing resources of the server are released to the greatest extent, and the method focuses on the face feature e