Face key point recognition and tracking method and system applied to cross-platform

Abstract

The invention relates to a face key point recognition and tracking method and system applied to cross-platform. The face key point recognition and tracking method comprises the steps: firstly collecting face images, marking the key point of each face image, and making a face image training sample set; based on a multi-task convolutional neural network algorithm, training to obtain a multi-task convolutional neural network model; collecting a face image, preprocessing the face image, loading a multi-task convolutional neural network model, reading a current frame image, and synchronously obtaining a face region and corresponding face key point position information; adopting the key point information of the first frame of face image as the input of the current frame, calculating the face keypoint information of the current frame through a multi-task convolutional neural network model, and judging whether the key point of the current frame of face image is in a successful tracking stateor not; and finally, after accumulatively tracking the predicted number of face key point information, calculating the Euler angle of the face through a pre-trained face deflection angle calculation model to complete face posture estimation.

Description

technical field [0001] The invention relates to the technical field of artificial intelligence, in particular to a cross-platform face key point recognition and tracking method and system. Background technique [0002] One of the existing face detection and tracking systems is based on traditional algorithms, implemented using open source libraries such as OpenCV or MTCNN, and the other is self-developed face key point detection and tracking by some companies, based on machine learning methods, using labeled data, training labeled data Data, get the method of the model to implement detection and tracking. [0003] Disadvantages of current technology: [0004] Algorithms that only rely on OpenCV or MTCNN open source libraries have disadvantages such as slow recognition speed and few key points to track, and third-party open source libraries have low security performance and many backdoors, which are easy to cause threats. In the face area detection of this type of system, t...

AI Technical Summary

Problems solved by technology

[0004] Algorithms that only rely on OpenCV or MTCNN open source libraries have shortcomings such as slow recognition speed and few key points to track, and third-party open source libraries have low security performance and many backdoors, which are easy to cause threats

In the face area detection of this type of system, the accuracy of face area detection is low due to problems such as the brightness, exposure, and contrast of the image in the video, or the face area detection takes a long time due to the size of the video image. The problem of inefficiency

[0005] Only based on the machine learning method, the number of recognized face key points is limited to a certain value due to the influence of the previous training model, and it is impossible to calculate more face key point data through the identified key points, which is inflexible and cannot be expanded. question

[0006] Based on the training algorithm of machine learning, the accuracy of the data set will affect the accuracy of the training model. At present, the original data set is manually marked in the industry, and the error rate is very large.

[0007] The current face recognition tracking system on the market will reduce the recognition rate by more than 50% and cannot achieve tracking in the case of different races, black skin, wearing glasses, hats, etc.

[0008] In the traditional tracking system, the method of improving robustness is limited to the use of Kalman filter, which will cause delays in the key points of the face recognized in the current frame after being processed by the Kalman filter.

Images