Face feature point tracking method and system, electronic equipment and storage medium

Abstract

The invention relates to a face feature point tracking method and system, electronic equipment and a storage medium, and the method comprises the steps: obtaining face feature points through an algorithm model, and predicting the key point position coordinates of a first frame of image through a key point detection model; obtaining a first key point box according to the key point position coordinates, and determining a second key point box at the same position of a second frame image based on the first key point box; inputting the second key point box into a regression network, and obtaining akey point coordinate offset of a second frame image through the regression network; through application of the method and the system, the problems that a tracking mode adopted by the method is relatively complex and a human face tracking algorithm model cannot be applied to a device end due to the fact that the model needs to use a large number of network parameters in related technologies are solved, and the purposes of reducing the network parameters required by the model and simplifying the tracking mode are achieved.

Description

technical field [0001] The present application relates to the technical field of image processing, in particular to a face feature point tracking method, system, electronic equipment and storage medium. Background technique [0002] With the development of Internet information technology and the significant improvement of computer software and hardware performance, the application of multimedia has become more and more popular. Among them, the face shape feature tracking technology refers to tracking one or more faces in a continuous video sequence, and The feature points of the face in each frame are output in real time. This technology can be applied in many scenarios, such as locking the activity track of a specific person, driver fatigue detection, processing of face frame images in mobile phone beautification, adding props in live video, etc. Solving the key point jitter problem in the video stream is the main problem [0003] In the related technology, a large amount...

AI Technical Summary

Problems solved by technology

[0003] In the related technology, a large amount of image data is output through the iterative calculation of the algorithm model or the convolutional neural network to output the facial feature points. Since the detection position of the frame before and after the video is different in this method, the key points detected in the video stream appear In addition, the tracking method adopted by this type of method is relatively complicated and requires a large number of tracking network parameters, which cannot be applied to the device side

[0004] At present, no effective solution has been proposed for the problems that the key points detected in related technologies are jittering, the tracking method is relatively complicated, and a large number of tracking parameters are required, which makes it impossible to apply to the device side.

Images