Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A Deep Learning-Based Air Target Tracking Method

An air target and deep learning technology, applied in the field of target tracking, can solve problems such as tracking drift

Active Publication Date: 2020-04-07
ZHONGBEI UNIV
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] In order to solve the tracking drift problem existing in tracking air targets based on convolutional neural network, the present invention proposes an air target tracking method based on deep learning

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
  • A Deep Learning-Based Air Target Tracking Method
  • A Deep Learning-Based Air Target Tracking Method
  • A Deep Learning-Based Air Target Tracking Method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0043] The air target tracking method based on deep learning includes the following steps:

[0044] 1. Build a multi-domain convolutional neural network (MDNet)

[0045] MDNet inputs a sample image of 107×107 pixels, including five hidden layers conv1, conv2, conv3, fc4 and fc5, and also includes a fc6, where conv1-3 is a convolutional layer, and fc4 and fc5 are fully connected layers ;The fully connected layer uses the relu activation function and dropout to prevent overfitting; the fc6 layer is a binary classification layer with K branches, each branch outputs a two-dimensional vector, and uses the softmax function to classify the target and background.

[0046] 2. Training bounding-box regression model in multi-domain convolutional neural network

[0047] (1) Calibrate the target position of the first frame

[0048] Calibrate the target position of the first frame as x 1 、y 1 is the coordinate value of the target position, s w , s h It is the width and height of the...

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 relates to an air target tracking method based on deep learning and specifically relates to the air target tracking method based on deep learning. The method comprises the following steps: through an MDNet, collecting positive samples of a first frame image and carrying out random sampling, and training a bounding-box model with the samples obtained after random sampling serving as adata set; then, training an AR model, order and parameters of which are determined by utilizing a minimum information criterion and a least square method, and estimating target motion trajectory andpredicating a target position; and then, serving the target position as a sampling center of the MDNet, and adjusting the target position through the bounding-box regression model, and thus tracking is finished accurately. The method can adaptively extract air target characteristics, combine with the AR model and effectively utilize motion information of a target, so that dependency of the MDNeton the target characteristics can be reduced greatly, pseudo-target interference is solved, and meanwhile, tracking precision is improved.

Description

technical field [0001] The invention relates to a target tracking method, in particular to an air target tracking method based on deep learning. Background technique [0002] Air target tracking is one of the key technologies of various detection systems such as aerospace. Due to the long observation distance, such targets often appear as small targets or even point targets due to the lack of shape and texture features during ground observation and tracking. , making detection and tracking difficult. Optical flow method, adjacent frame difference method and background subtraction are the main methods currently used, but they are often only for specific targets and need to select a suitable tracking algorithm based on prior knowledge, which is not conducive to engineering applications. [0003] Recently, deep learning has successfully broken through the constraints of fixed state models in many fields such as image classification and target detection. Both the neural networ...

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 Patents(China)
IPC IPC(8): G06T7/246G06N3/04G06N3/08
Inventor 蔺素珍郑瑶任之俊
Owner ZHONGBEI UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com