Human body posture estimation optimization method
An optimization method and a technology of human body posture, applied in neural learning methods, calculations, computer components, etc., can solve problems such as redundancy and insufficient accuracy of human body area frames, and achieve the effect of improving robustness
Inactive Publication Date: 2020-11-10
NANJING UNIV OF POSTS & TELECOMM
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] The purpose of the present invention is to provide a human body posture estimation optimization method to solve the problem of insufficient accuracy and redundancy of human body area frames in the case of multi-person key point detection or background crowding
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
[0020] The present invention will be further explained below in conjunction with the accompanying drawings and specific implementation examples. It should be understood that these examples are only used to illustrate the present invention and not limit the scope of the present invention.
[0021] Such as figure 1 Shown, a kind of human body pose estimation optimization method of the present invention comprises the following steps:
[0022] First, input a human body area frame dataset preprocessed by Faster-RCNN, here only focus on its output, the area frame is represented by point coordinate information: A=(L,P), where L represents the label of the area frame, P denotes the set of point coordinates forming the region box. Use the space transformation network to extract a high-quality region box from the obtained area box. The local network in the space transformer is used to regress the transformation parameter θ, and the space transformation function is a two-dimensional aff...
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 human body posture estimation optimization method. The method comprises the steps: 1, obtaining a human body area frame through preprocessing of a target detection network; 2, reducing errors existing in the human body area frame obtained in the step 1 through a space transformation network, and removing redundancy through a non-maximum suppression algorithm; 3, estimating key point positions in the human body region frame through a deep high-resolution network; and step 4, converting into an original spatial state through a spatial inverse transformation network, andmapping the positioned key point positions to full resolution. According to the method, target detection is combined, and for the condition that the convolutional neural network is limited by spatialinvariance of input data, a spatial transformation network is introduced to realize spatial feature mapping conversion. Besides, for a partial region frame redundancy phenomenon, a non-maximum suppression algorithm is introduced, the optimized human body region frame is used to construct a deep high-resolution network prediction heat map, and the robustness of the model is improved on the basis of the original model.
Description
technical field [0001] The invention relates to an optimization algorithm for human body posture estimation based on a deep high-resolution network and a space transformation network, and belongs to the technical field of computer vision posture estimation. Background technique [0002] Human key point detection refers to the process of estimating and detecting the key points of tasks in film and television videos, which plays a fundamental role in the research of computer vision fields such as human-computer interaction and behavior recognition. Due to the complexity of human body posture, the visibility of its key points is greatly affected by occlusion, viewing angle, etc. Human key point detection is also a challenging topic in computer vision. [0003] This topic requires that the human body key points in the video frames be located and output, such as the head, neck, shoulders, etc., by using the detected and tracked character area frames in the continuous video frames...
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): G06K9/00G06K9/46G06N3/08G06T3/00G06T3/40
CPCG06T3/4053G06N3/084G06V40/103G06V10/462G06V2201/07G06T3/02
Inventor 张怡静陈志张懿扬赵彤彤岳文静李玲娟
Owner NANJING UNIV OF POSTS & TELECOMM