1）人脸采集模块，人脸图像的来源可以是来自于本地文件、网络URL或是视

1) Face acquisition module. The source of face images can be from local files, network URLs or video frames. This time the design uses OpenCV functions to obtain from files and cameras. <br>2) Face correction module, because the intercepted face image is likely to be crooked, such as tilting the head, in order to improve the recognition accuracy, an affine transformation method provided in OpenCV needs to be used, using the key of the detected face Click to correct the intercepted face to a standard position. <br>3) Face detection module, which is used to preprocess the input image, determine the number of faces in the picture, and detect the location of the face and the coordinates of the key points of the face, and remove the face image from the background Intercepted down. At the same time, it should maintain good robustness and real-time detection performance in real scenes. <br>4) Face recognition module. The face recognition module generally consists of two parts, namely, facial feature extraction and the establishment of a face index database. After getting the pre-processed face image, input the face into the pre-trained model, extract 512-dimensional face features, and build an index for it. <br>5) The face attribute module first preprocesses the face image to be detected to obtain a more standard face image.

1) Face acquisition module, the source of face images can be from local files, network URLs or video frames, this design uses OpenCV functions from files and cameras.<br>2) Face correction module, because the intercepted face image is likely to be crooked, such as crooked head, in order to improve the accuracy of recognition, it is necessary to use open CV to provide an affine transformation method, the use of detected face key points to intercept the face corrected to a standard position.<br>3) Face detection module, in which the module is used to pre-process the input image, determine the number of faces in the picture, as well as detect the location of the face and the key point coordinates of the face, and the face image from the background. At the same time, in the real-world scenario should maintain a good robustness and real-time detection performance. <br>4) Face recognition module, face recognition module is generally composed of two parts, that is, face feature extraction and the establishment of face index library. After obtaining the preprocessed face image, the face is entered into the pre-trained model, the 512-dimensional face feature is extracted and indexed.<br>5) Face attribute module, first of all, the detection of face image pre-processing, to get a more standard face image.

1) Face acquisition module, the source of face image can be from local files, network URL or video frames, this design uses opencv function to obtain from files and cameras.<br>2) In face correction module, because the intercepted face image is likely to be crooked, such as crooked head, in order to improve the recognition accuracy, we need to use an affine transformation method provided by OpenCV to correct the intercepted face to a standard position by using the detected face key points.<br>3) Face detection module, which is used to preprocess the input image, judge the number of faces in the image, detect the position of the face and the coordinates of the key points of the face, and intercept the face image from the background. At the same time, it should keep good robustness and real-time detection performance in the real scene.<br>4) Face recognition module, face recognition module is generally composed of two parts, namely face feature extraction and face index database. After the preprocessed face image is obtained, the face is input into the pre trained model, and 512 dimensional face features are extracted and indexed.<br>5) In the face attribute module, a standard face image is obtained by preprocessing the detected face image<br>