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

A UAV small target detection method based on motion features and deep learning features

A technology of small target detection and deep learning, applied in the field of small target detection of drones, can solve problems such as small size and difficult detection of drones

Active Publication Date: 2021-03-30
CHONGQING UNIV OF POSTS & TELECOMM
View PDF1 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

UAVs are smaller in size than pedestrians, airplanes, vehicles, etc. Especially in long-range imaging, the size of UAVs is very small, which makes vision-based UAV detection more difficult

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 UAV small target detection method based on motion features and deep learning features
  • A UAV small target detection method based on motion features and deep learning features
  • A UAV small target detection method based on motion features and deep learning features

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0077] The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0078] In the present invention, the candidate target detection module based on motion features performs video image stabilization on the original video, and extracts the moving target area in the video through low-rank matrix analysis;

[0079] Candidate target detection module based on deep features extracts candidate targets from video images through an improved region generation network model;

[0080] The improved candidate area generation network is based on the traditional area generation network, which modifies the network structure and the scale of the candidate area, and replaces the network layer that outputs the feature map;

[0081] The candidate region fusion module is to fuse the candidate regions obtained in steps S2 and S3;

[0082] The candidate target recognition module based on the dual-channel deep neural network is to...

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 a small target detection method of an unmanned aerial vehicle based on motion features and deep learning features, and belongs to the technical field of image processing and computer vision. First, the input video data set is processed by the video image stabilization algorithm to compensate for camera motion; the motion candidate target area is detected in the image analysis; the video data set is divided into two parts, and the training data set is used to train the improved candidate area to generate a network model; Through the candidate area generation network based on deep features obtained through training, the candidate target is generated from the video image of the test set through the network; the candidate target area is fused; the training data set is used to train the model based on the dual-channel deep neural network, and Apply the model to get the recognition result. The target tracking method based on multi-layer depth features is applied to the recognition results of the previous step to obtain the final location of the UAV. The invention can accurately detect the unmanned aerial vehicle in the video image, and provides support for subsequent research in the field of intelligent monitoring of the unmanned aerial vehicle.

Description

technical field [0001] The invention belongs to the technical field of image processing and computer vision, and relates to a small target detection method for drones based on motion features and deep learning features. Background technique [0002] Now, with the availability and sophistication of commercial drones skyrocketing, drone sales have multiplied, and drones flying in the public domain have become commonplace. UAVs not only appear in the scenes of popular variety shows and romantic proposal ceremonies, but also can spray pesticides over farmland, replace workers in high-altitude cleaning operations, and be used for surveying and mapping, forest fire prevention, military reconnaissance, etc. However, with the rapid development of drones, dangerous accidents caused by drones are also increasing, posing threats to public safety, privacy leaks, and military security. [0003] In recent years, UAV detection technology can be roughly divided into acoustic detection (Aco...

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): G06T7/246G06N3/04
CPCG06T7/246G06T2207/10016G06T2207/20081G06N3/045
Inventor 高陈强杜莲王灿冯琦汤林汪澜
Owner CHONGQING UNIV OF POSTS & TELECOMM