Gait data set synthesis method oriented to complex scene fine-grained attribute driving

A complex scene and data collection technology, applied in the field of computer vision, can solve problems such as difficulty in collecting data, large differences, and low quality of data sets

Active Publication Date: 2021-05-07
ZHEJIANG UNIV +1
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

There are currently two key points in this task: the first is how to solve the problem that the existing gait recognition data sets are of low quali

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
  • Gait data set synthesis method oriented to complex scene fine-grained attribute driving
  • Gait data set synthesis method oriented to complex scene fine-grained attribute driving
  • Gait data set synthesis method oriented to complex scene fine-grained attribute driving

Examples

Experimental program
Comparison scheme
Effect test

preparation example Construction

[0040] refer to figure 1 and figure 2 , in a preferred embodiment of the present invention, a fine-grained attribute-driven gait data set synthesis method for complex scenes is provided, including the following steps:

[0041] S1. Generate several multi-attribute pedestrian 3D models.

[0042] In this embodiment, the specific implementation method of step S1 is:

[0043]Use the MakeHuman tool to randomly generate N 3D character models, each character model has several random character attributes; all 3D character models form a pedestrian 3D model group where M i is the i-th multi-attribute pedestrian 3D model, and N is the total number of models, which is set according to the actual situation.

[0044] Wherein, the random character attributes may include gender, age, weight, ethnicity, etc., and may also include other character attributes, such as height, body shape, and the like.

[0045] S2. Obtain several groups of pedestrian gait movements for binding.

[0046] In ...

Embodiment

[0064] The implementation method of this embodiment is as described above, and the specific steps will not be described in detail. The following only shows the effect of the case data.

[0065] In this embodiment, based on the fine-grained attribute-driven gait data set for complex scenes established by S1-S5, the effect of the generated virtual gait data set is tested by pre-training with the mainstream gait recognition framework. The present invention is tested on two data sets with true value labels, that is, the virtual gait data set D generated by this method is used as a pre-training data set, and the CASIA-B data set is used as a test data set. In the GaitSet and GaitPart methods verified on . see Figure 4 As shown, the data set is as follows:

[0066] Virtual data set: the virtual data set generated by the method of the present invention includes 11268 IDs, 33 perspectives, and 1031844 gait sequences. The data comparison of other mainstream gait data sets in this da...

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 discloses a gait data set synthesis method oriented to complex scene fine-grained attribute driving. A group of virtual gait data sets capable of effectively improving gait recognition performance is generated by using the method. The method specifically comprises the following steps: generating a multi-attribute pedestrian 3D model; acquiring a pedestrian gait action for binding; binding the gait actions to the generated 3D model to obtain a 3D pedestrian model of walking; building a multi-angle gait data capturing scene in the 3D engine; collecting gait silhouette image data and fine-grained labels, and carrying out data preprocessing; and pre-training and checking the effect of the generated virtual data set by using a mainstream gait recognition framework. The method is used for generating a virtual pedestrian gait data set, the generated data has the characteristics of high quality, multiple views, multiple attributes and the like, and the recognition performance and robustness are well improved by using the data set for pre-training in a mainstream algorithm.

Description

technical field [0001] The invention belongs to the field of computer vision, and in particular relates to a fine-grained attribute-driven gait data set synthesis method for complex scenes. Background technique [0002] Pedestrian gait recognition is defined as the following problem: in a set of pictures or video sequences containing pedestrian walking gestures, identify the identity of pedestrians by the way and style of walking. In recent years, gait recognition has been widely used in smart cities, smart security and other fields. There are currently two key points in this task: the first is how to solve the problem that the existing gait recognition data sets are of low quality and the data collection is difficult; the second is how to solve the problem of the real scene in the real gait data set Questions that vary widely. For the first point, the present invention believes that the existing gait data sets often face the problems of high collection cost and low qualit...

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
CPCG06V40/25G06F18/214
Inventor 李玺赵彧涵窦洹彰张文虎张芃怡董霖方毅
Owner ZHEJIANG UNIV
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