Face detection alignment network knowledge distillation method and device
A technology of face detection and distillation method, which is applied in the field of face recognition, can solve the problems of poor effect of binary classification target detection and high difficulty of distillation, and achieve the effect of improving the final accuracy
Pending Publication Date: 2022-02-18
XIAMEN MEIYA PICO INFORMATION
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
The particularity of face detection / alignment network knowledge distillation is that the face detection / alignment network includes multiple tasks such as face classification / face frame regression / key point detection. less effective
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
[0035] In order to make the purpose, technical solutions and advantages of the present disclosure clearer, the present disclosure will be further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.
[0036] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in one or more embodiments of the present disclosure shall have the usual meanings understood by those skilled in the art to which the present disclosure belongs. "First", "second" and similar words used in one or more embodiments of the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. "Comprising" or "comprising" and similar words mean that the elements or items appearing before the word include the elements or items listed after the word and their equivalents, without excluding other elements or items. Words such as "connected" or "connecte...
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 provides a face detection alignment network knowledge distillation method and device, and the method comprises the steps: 1, carrying out the training to obtain a first detection alignment network meeting the preset face detection alignment precision, and enabling the first detection alignment network to have a first output; 2, constructing a second detection alignment network having the same task head as the first detection alignment network, wherein the second detection alignment network has a second output; 3, determining a real label; 4, constructing an overall loss function of the distillation model based on the first output of the first detection alignment network, the second output of the second detection alignment network and a real label; and 5, updating parameters of the second detection alignment network based on the overall loss function until corresponding convergence. And the detection precision of the existing face detection network is improved.
Description
technical field [0001] One or more embodiments of the present disclosure relate to the technical field of face recognition, and in particular, to a knowledge distillation method and device for a face detection alignment network. Background technique [0002] As the basic task of face recognition, face detection / alignment is crucial to improving the performance of the entire face recognition system. Face detection / alignment algorithms for actual business needs need to achieve fast and accurate face detection on limited computing resources , thus placing a requirement on model compression. Common model compression techniques include model pruning, model quantization, and knowledge distillation, among which knowledge distillation is a lossless model compression technique. The core idea of knowledge distillation is to teach a smaller student network model through a larger teacher network model. The goal is to let the student network learn the dark knowledge in the teacher net...
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 Applications(China)
IPC IPC(8): G06V40/16G06K9/62G06N3/04G06N3/08G06V10/774
CPCG06N3/04G06N3/082G06F18/214
Inventor 钟敏毕永辉蓝坤宏潘锟欧阳天陈毓靖
Owner XIAMEN MEIYA PICO INFORMATION