Multi-task cascade face aligning method based on deep learning

A face alignment and deep learning technology, applied in neural learning methods, instruments, biological neural network models, etc., can solve problems such as facial expression, posture, occlusion and gender robustness

Inactive Publication Date: 2017-08-11
四川云图睿视科技有限公司
View PDF3 Cites 59 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0003] Existing face alignment methods are not robust to facial expression, pose, occlusion, and gender

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
  • Multi-task cascade face aligning method based on deep learning
  • Multi-task cascade face aligning method based on deep learning
  • Multi-task cascade face aligning method based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0019] A multi-task cascaded face alignment method based on deep learning, such as figure 1 shown, including the following steps:

[0020] Step 1. Sample collection

[0021] (1) Manually mark the key points of the face, such as figure 2 shown;

[0022] (2) Mark the face attribute label:

[0023] Gender label: male is 0, female is 1; label whether to smile: 1 if smiling, 0 if not smiling; label whether to wear glasses: 1 if wearing glasses, otherwise 0; attitude label: Roll, Pitch, Yaw are all -90 to 90 range. The face pose is the right-hand Cartesian coordinate in 3D space. Pitch is a rotation around the X axis, also called pitch angle. yaw is the rotation around the Y axis, also called the yaw angle. Roll is a rotation around the Z axis, also called roll angle.

[0024] Step 2. Sample preprocessing

[0025] (1) The face image is grayscaled and normalized:

[0026] The face image uses a grayscale image, and the image is normalized, that is, the mean value of the ima...

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 multi-task cascade face aligning method based on deep learning. The method comprises the following steps: employing the convolutional neural network to perform face aligning model training, wherein a main line network comprises 8 convolution layers, 4 normalization layers, 4 activation layers, 4 pooling layers and a full connection layer, and the structure comprises each two convolution layers, then one normalization layer, one activation layer and one pooling layer, and so on; and behind the first three normalization layers and the last one full connection layer, the full connection layer is taken as a branch network for prediction to predict face key points and face attributes. The multi-task cascade face aligning method based on the deep learning adds auxiliary information such as genders, smile or not, having eyes or postures or not in the model training process to realize multi-task learning, and there is prediction output at each two layers of the network so as to realize coarse-to-fine cascade face key-point location detection and improve the robustness of face expression, posture, gender and shielding through face aligning.

Description

technical field [0001] The invention relates to a multi-task cascaded face alignment method based on deep learning. Background technique [0002] Face alignment is applied in a wide range of fields: 1. Face recognition, key point prediction is an extremely important link in face recognition. The robustness of face alignment is directly related to the accuracy of face recognition; 2. Facial organs Positioning, organ tracking. Through face alignment, we can locate each part of the face and extract the corresponding part features for subsequent development; 3. Expression recognition. After face alignment, we can use the aligned face shape to analyze the expression state of the face; 4. Face caricature / sketch image generation. After face alignment, we can generate face cartoons and sketches; 5. Virtual reality and augmented reality. After face alignment, we can make many interesting applications; 6. Face aging, rejuvenation, and age inference. Feature Fusion / Image Enhancemen...

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): G06K9/00G06N3/08
CPCG06N3/08G06V40/165
Inventor 刘云楚
Owner 四川云图睿视科技有限公司
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

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap