A face model accuracy correction method, device and storage medium
A face model, the technology of the first face, applied in the field of face detection, can solve the problems of low re-existence efficiency, reduced face recognition accuracy, and large amount of training
Active Publication Date: 2022-07-15
XIAMEN MEIYA PICO INFORMATION
View PDF4 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] In view of the above-mentioned face detection algorithm replacement in the face recognition system, the face recognition accuracy will be greatly reduced, the re-existence efficiency is low, and the amount of training is large.
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0045] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.
[0046]figure 1 An exemplary apparatus architecture 100 to which the face model accuracy correction method or the face model accuracy correction apparatus according to the embodiments of the present application can be applied is shown.
[0047] like figure 1 As shown, the apparatus architecture 100 may include terminal devices 101 , 102 , 103 , a network 104 and a server 105 . The network 104 is a medium used to provide a communi...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a face model accuracy correction method, device and storage medium, respectively adopting a first face recognition system including a first face detection algorithm and a second face recognition system including a second face detection algorithm The system extracts face images from the same data set to obtain target sample pairs; uses the face recognition training data set as the training set to extract the facial features in the first face recognition system by fine-tuning the hyperparameters and executing the data enhancement transformation scheme The model is trained; the first face recognition system is used as the student network, and the facial feature extraction model after training in the above steps is the teacher network, and the target sample is used to train the student network through knowledge distillation, and the knowledge distillation adopts heterogeneous homology. The sample self-distillation structure; the second face detection algorithm is combined with the student network trained in the above steps to form a third face recognition system. The method is high-speed and effective, and simplifies the whole process of accuracy correction after the face detection algorithm is replaced.
Description
technical field [0001] The invention relates to the field of face detection, in particular to a face model accuracy correction method, device and storage medium. Background technique [0002] With the continuous emergence of new face detection technologies, how to replace face detection algorithms in face recognition systems with high efficiency and quality has become an urgent problem. Since the face feature extraction model is trained and implemented on the basis of the sample set extracted by the face detection algorithm, it is very dependent on various parameters set during face detection, and subtle differences in position and angle will lead to later face recognition. A substantial drop in accuracy. The traditional accuracy correction method is to use the new face detection algorithm to re-establish the face feature training sample set and re-adjust the model, which is very inefficient when facing hundreds of millions of face image libraries. [0003] The traditional...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Patents(China)
IPC IPC(8): G06V40/16G06V10/25G06V10/82G06T3/00G06T5/00G06N3/04G06N3/08
CPCG06N3/08G06T2207/10004G06T2207/20081G06T2207/20084G06T2207/20104G06T2207/30201G06V40/161G06V40/168G06V10/25G06N3/045G06T3/147G06T5/92
Inventor 高志鹏黄仁裕姚灿荣赵建强张辉极杜新胜江冠南
Owner XIAMEN MEIYA PICO INFORMATION