Unlock instant, AI-driven research and patent intelligence for your innovation.

SSD face detection method based on deep learning

A face detection and deep learning technology, applied in the field of deep learning, can solve the problems of inaccurate position prediction and unsatisfactory effect of small objects, etc., and achieve the effects of fast detection speed, improved efficiency and high precision

Inactive Publication Date: 2018-07-06
上海玄彩美科网络科技有限公司
View PDF0 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

At present, the more efficient face detection algorithm based on deep learning, YOLO (You Only Look Once), although the detection speed of objects is very fast, and it is completely real-time, but its position prediction is not accurate enough, and the effect on small objects is not ideal.

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 more

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0029] The present invention will be further described below.

[0030] A face detection method based on the SSD of deep learning, comprising the following steps:

[0031] 1) Input image.

[0032] The input images are classified as square or oblong rectangles. If it is a square, proceed directly to step 2); if it is a rectangle, take multiple square subimages from the image, and perform step 2) for each subimage.

[0033] 2) Resize the obtained picture to 300*300 and send it to the network.

[0034] 3) Perform multi-level feature map extraction.

[0035] Extract the feature maps of the six layers conv4_3, fc7, conv8_2, conv9_2, conv10_2, pool11 respectively.

[0036] 4) Each location in each feature map, each location corresponds to multiple defaultboxes.

[0037] Since for each box in the k boxes at a position of a feature map cell in the feature map, we need to calculate c classes, the score of each class, and the 4 deviations of this box relative to its default box Off...

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

No PUM Login to View More

Abstract

The invention provides an SSD face detection method based on deep learning. The method comprises the steps that (1) an image is input; (2) an acquired picture is resizes into 300*300 and sent into a network; (3) multi-level feature map extracting is carried out; (4) at each location in each feature map, each location corresponds to multiple default boxes; (5) for the characteristics of all defaultboxes, the confidence coefficient of the category thereof and the relative offset of the default boxes are regressed respectively; (6) for each default box, the class score of each category is regressed; (7) a default box is selected according to the class score of the face category in regressed categories; and (8) according to the prediction result of the default box, the tilt angle of a face can be known, and the orientation of the face is known.

Description

technical field [0001] The invention belongs to the field of deep learning, and in particular relates to an SSD (Single Shot Multi-boxes Detector, single shot multi-boxes detection) face detection method. Background technique [0002] Face detection is a computer technology that finds the position and size of a human face in any digital image, and its research has important academic value. Human face is a kind of natural structural target with quite complex detail changes. The challenge for this type of target is: human face has the variability of patterns due to different appearance, expression, skin color, etc.; in general, human face, There may be accessories such as glasses and beards; as a three-dimensional object, the face image is inevitably affected by the shadow produced by the light. The purpose of face detection is to judge whether there is a human face in the input face image or video frame image, if there is a human face, determine the position of the human fac...

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
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06K9/00
CPCG06V40/172G06V40/161G06V40/168
Inventor 金海强朱毅李汉曦钱胜
Owner 上海玄彩美科网络科技有限公司