Face key point detection method, device and equipment and storage medium

Abstract

The embodiment of the invention discloses a face key point detection method, device and equipment and a storage medium. The method comprises the steps of obtaining a face original image of a current to-be-detected frame, and obtaining posture correction information of a face in the face original image of the current to-be-detected frame; correcting the face posture in the original face image according to the posture correction information to obtain a face correction image; carrying out face key point detection on the face correction image by adopting a key point detection network model to obtain correction key points; and performing reverse posture correction on the correction key points according to the posture correction information to obtain target face key points of the original face image. By utilizing the method, compared with the existing detection method, the detection precision of key point detection on the special face image can be ensured under the condition that the scale of the key point detection network model is not increased, and meanwhile, the increase of the detection time of the key point detection is avoided, so that the detection effect of real-time performanceand accuracy is achieved.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of image recognition, and in particular, to a face key point detection method, device, equipment, and storage medium. Background technique [0002] In the field of computer vision, face key point detection has been widely concerned by academia and industry. Its main function is to accurately locate the key points on the face (such as eyes, nose, mouth corners, face contour points, etc.), To prepare for subsequent face image processing (such as face alignment and face recognition), face key point detection has played an important role in application scenarios such as biometric information verification, surveillance security, and live video. [0003] At present, the neural network model is mostly used to realize face key point detection. In order to ensure the accuracy of the detection results, a large-scale neural network model is often used. The size of the neural network model will be...

AI Technical Summary

Problems solved by technology

[0003] At present, the neural network model is mostly used to realize face key point detection. In order to ensure the accuracy of the detection results, a large-scale neural network model is often used. The size of the neural network model will be limited, which will cause the execution terminal to perform face keying on images containing faces with large angles and large postures (head up at 45 degrees, head down at 45 degrees, head tilted to 90 degrees, etc.) Point detection does not perform well

[0004] In order to improve the key point detection effect of faces with large angles and large poses, the usual method is to add more face training samples with large angles and large poses to train the current neural network model, but in actual operation , such face training samples with large angles and large poses are not easy to obtain, and under the premise of limiting the network scale, the increase in sample complexity will often lead to a decrease in the accuracy of the neural network model in detecting normal face images. In addition, , if the network scale is increased to ensure the detection accuracy, the detection time of the image frame will be increased, and the real-time requirements of the detection cannot be guaranteed.

Images