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

Method for simultaneously finishing pedestrian detection and pedestrian re-identification

A pedestrian re-identification and pedestrian detection technology, applied in the field of deep learning, can solve problems such as data asymmetry, and achieve the effect of improving efficiency, efficient real-time processing, and improving classification and recognition capabilities.

Inactive Publication Date: 2019-05-14
CHINA CHANGFENG SCI TECH IND GROUPCORP
View PDF0 Cites 12 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In view of the above defects of the prior art, the present invention proposes a method for simultaneously completing pedestrian detection and pedestrian re-identification based on deep learning. problem, improve the accuracy of pedestrian re-identification technology

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
  • Method for simultaneously finishing pedestrian detection and pedestrian re-identification
  • Method for simultaneously finishing pedestrian detection and pedestrian re-identification
  • Method for simultaneously finishing pedestrian detection and pedestrian re-identification

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0019] figure 1 Shown is the structure diagram of the core deep convolutional neural network of the invention.

[0020] The main network structure adopted by the present invention is VGG16 network + PPN area generation network + fully connected identification layer, which mainly includes a large number of convolutional layers, pooling layers and fully connected layers. The model uses the convolutional network to learn the hidden discriminative features in the picture to overcome the artificial design interference of traditional features, in which the PPN pedestrian candidate area network and its corresponding target regression function help generate high-probability pedestrian frames. The ROI-pooling layer solves the problem of different sizes of pedestrian frame extraction feature maps, which is similar to the resize function. The pedestrian position regression target at the end further corrects the position of the pedestrian frame, and the output of the fully connected laye...

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 a method for simultaneously finishing pedestrian detection and pedestrian re-identification. The method comprises the following steps of: extracting video frames in preset areasunder cameras at different angles, and manually calibrating pedestrian position frames and related label information to form training data; first five convolutional layers of a VGG16 convolutional neural network structure are adopted as a basic network, then a local pedestrian candidate network PPN is added to generate candidate pedestrian frame positions, and ROI-pooling is carried out accordingto a PPN network output result; carrying out pooling operation, and carrying out feature fusion by adopting three full connection layers; Taking the output of the last full connection layer as a feature representation, and establishing a feature dictionary-feature retrieval library is used for matching the deep learning features of the pedestrian region part determined by the detection model withthe similarity of all the pedestrian special symptoms in the feature retrieval library; and when the similarity of the two features meets a preset requirement, determining that the pedestrian in thetest picture and the person with the maximum similarity in the picture library are the same.

Description

technical field [0001] The invention belongs to the technical field of pattern recognition, and more specifically, relates to a deep learning method for pedestrian detection and pedestrian re-identification. Background technique [0002] Pedestrian re-identification refers to the technology of automatically matching the same pedestrian object under different viewing angles of non-overlapping cameras. face and other specific information to directly find the same target. The re-identification technology requires information such as pedestrian appearance, contour texture, etc., to complete the recognition and matching under the appropriate feature space and measurement criteria. This task first requires pedestrian detection to select a high-probability pedestrian frame, then performs feature extraction and similarity matching on multiple candidate frames, and finally locks the retrieval target. [0003] Deep learning is the new darling of video image processing tasks. With th...

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): G06K9/00G06K9/62G06N3/08
Inventor 单鼎一刘惟锦张晓林
Owner CHINA CHANGFENG SCI TECH IND GROUPCORP