Face attribute identification method and device and model establishing method

An attribute recognition and attribute technology, applied in the field of image recognition, can solve the problems of face attribute recognition method dependence, difficulty, learning multiple attributes, etc.

Inactive Publication Date: 2017-09-08
深圳市深网视界科技有限公司
View PDF7 Cites 25 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0008] In order to overcome the deficiencies of the prior art, one of the purposes of the present invention is to provide a face attribute recognition method, which can solve the problem that the existing face attribute recognition method relies on accurate key point positioning of training and test data sets and artificial Define the associated area of ​​the attribute, and it is difficult to learn multiple attributes through a unified model
[0009] The second object of the present invention is to provide a method for establishing a face attribute recognition model, which can solve the problem tha...

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

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Face attribute identification method and device and model establishing method
  • Face attribute identification method and device and model establishing method
  • Face attribute identification method and device and model establishing method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0064] figure 1 It is a schematic flow chart of the face attribute recognition method in Embodiment 1 of the present invention; figure 2 yes figure 1 The schematic diagram of the process of the facial attribute recognition method, those skilled in the art should understand, figure 2 The shared convolutional network, the first convolutional layer and the second convolutional layer in can each include one or more convolutional layers, which can perform multiple operations on corresponding inputs.

[0065] Such as figure 1 with 2 Shown, a kind of face attribute recognition method of the present embodiment, it comprises the following steps:

[0066] Step S110, acquiring the current response area from the current image.

[0067] Further, the obtaining the current response area from the current image specifically refers to obtaining the current response area from the current image through a convolutional neural network, and the convolutional neural network includes a convolut...

Embodiment 2

[0090] Such as image 3 A method for establishing a human face attribute recognition model shown, Figure 4 It is a schematic diagram of the structure of a face attribute recognition model. The face attribute recognition model establishment method comprises the following steps:

[0091] Step S210, establish a convolutional neural network, the convolutional neural network includes a shared convolutional network, a first task branch and a second task branch; the shared convolutional network is used to process the current image, and the first task The branch is used to obtain the current response area from the processed current image, and calculate the attribute association area according to the average response area and the current response area, and the second task branch is used to make an interest area for the attribute association area Pooling to obtain a pending feature map of a preset size, and predict face attributes based on the pending feature map.

[0092] The share...

Embodiment 3

[0117] Such as Figure 5 The shown face attribute recognition device includes:

[0118] An acquisition module 110, configured to acquire the current response area from the current image;

[0119] Further, the acquisition module 110 includes:

[0120] a first calculation unit, configured to calculate an initial feature map of the current image;

[0121] a second calculation unit, configured to calculate a response map of the initial feature map;

[0122] an extraction unit, configured to extract the current response area according to the response map;

[0123] A calculation module 120, configured to calculate an attribute association area according to the average response area and the current response area;

[0124] A region module 130, configured to perform region-of-interest pooling on the attribute-associated region to obtain a pending feature map of a preset size;

[0125] Further, the area module 130 includes:

[0126] a clipping unit, configured to clip a region of ...

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 present invention discloses a face attribute identification method and device and a model establishing method, wherein the face attribute identification method comprises the following steps of obtaining a current response area from a current image; according to an average response area and the current response area, calculating an attribute association area; carrying out the area-of-interest pooling on the attribute association area to obtain an undetermined feature graph of a preset size; according to the undetermined feature graph, predicting the face attribute. The face attribute identification method can deform and pool the attribute association area into the undetermined feature graph of the preset size by the area-of-interest pooling, and solves the problem that the scales of the association areas corresponding to the conventional attributes are different, so that a subsequent uniform model-based multi-attribute identification problem is disadvantageous, thereby realizing the face image multi-attribute identification by a convolutional neural network.

Description

technical field [0001] The invention relates to an image recognition technology, in particular to a face attribute recognition method, device and model building method. Background technique [0002] Face attribute recognition refers to the estimation, discrimination and analysis of the attributes shown in the face image. Face attributes include gender, age, expression, movement, whether to wear glasses, whether to wear sunglasses, eye size, eyes open or closed, mouth open or closed, hair length or straight or curly, front or side, etc. It can be seen that the attributes of the face can be divided into long-term attributes and short-term attributes. Long-term attributes refer to attributes that will not change for a period of time, such as gender, age, etc. Short-term attributes refer to attributes attached to the face that can be removed at any time, such as wearing masks, glasses, necklaces, makeup, hairstyles, etc. [0003] At present, the methods of face attribute recog...

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
IPC IPC(8): G06K9/00G06K9/62
CPCG06V40/16G06V40/171G06V40/178G06F18/214
Inventor 徐静童长毅赵瑞
Owner 深圳市深网视界科技有限公司
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products