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

A real-time target detection method and device for UAV airborne platform deployment

A target detection and unmanned aerial vehicle technology, applied in the field of computer vision and neural network optimization, can solve the problems of unable to meet real-time requirements, huge network scale, etc., achieve the effect of reducing model size, improving detection accuracy, and improving accuracy

Active Publication Date: 2022-03-08
SUN YAT SEN UNIV
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the existing deep learning-based target detection algorithms still have the following defects: For example, the Fast-cnn algorithm of the RCNN series cannot meet the real-time requirements even under the powerful GPU computing resources on the ground, and the YOLO series, which is superior in speed However, because it requires a huge network scale, it cannot be deployed on embedded airborne platform devices

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 real-time target detection method and device for UAV airborne platform deployment
  • A real-time target detection method and device for UAV airborne platform deployment
  • A real-time target detection method and device for UAV airborne platform deployment

Examples

Experimental program
Comparison scheme
Effect test

no. 1 example

[0048] see Figure 1-4 .

[0049] like figure 1 As shown, the present embodiment provides a real-time target detection method for UAV airborne platform deployment, at least including the following steps:

[0050] S1. Deploy the pre-built LiteDenseHG-Net network model to the UAV's airborne platform, and collect the ground image in real time through the UAV's onboard camera.

[0051] Specifically, for step S1, first deploy the LiteDenseHG-Net network model on the onboard GPU platform, and capture ground-facing images through the onboard camera.

[0052] Among them, the LiteDenseHG-Net network model used in this embodiment is an ultra-lightweight network structure oriented to platforms with limited computing resources. The LiteDenseHG-Net network is very small. In the input part, in order to strengthen information between layers Multiplexing of streams, making full use of shallow layer information, this embodiment introduces a deformed dense connection from a certain layer to ...

no. 2 example

[0104] see Figure 5 .

[0105] like Figure 5 As shown, the present embodiment provides a real-time target detection device for unmanned aerial vehicle platform deployment, including:

[0106] The deployment module 100 is used to deploy the pre-built LiteDenseHG-Net network model to the airborne platform of the UAV, and collect the ground image in real time through the onboard camera of the UAV.

[0107] Specifically, for the deployment module 100, it is used to deploy the LiteDenseHG-Net network model on the onboard GPU platform, and capture and collect ground images through the onboard camera.

[0108] The image preprocessing module 200 is configured to perform image preprocessing on the ground image captured by the airborne camera in real time, and store it in the database of the airborne platform.

[0109] Specifically, for the image preprocessing module 200, it is used to resize all the collected images to a uniform size of 320*240; for a UAV RGB three-channel image c...

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 real-time target detection method and device for deployment on an unmanned aerial vehicle platform. The method includes: deploying a pre-built LiteDenseHG‑Net network model to the unmanned aerial platform, The onboard camera of the aircraft collects the ground image in real time; after the image preprocessing is performed on the image of the ground image collected by the airborne camera in real time, it is stored in the database of the airborne platform; the image after the image preprocessing is input to the The LiteDenseHG-Net network model is forward calculated to obtain the corresponding target detection result; the target detection result is sent to the control unit of the unmanned aerial vehicle in real time, so that the control unit can perform the target detection according to the target detection result. Human-machine control in real time. The present invention carries the LiteDenseHG-Net network model on the UAV airborne platform, realizes the purpose of real-time detection of the target on the platform with limited computing resources, improves the detection accuracy, and accurately detects the UAV in real time after the detection is completed. real-time control.

Description

technical field [0001] The invention relates to the technical field of computer vision and neural network optimization, in particular to a real-time target detection method and device for deployment on an unmanned aerial vehicle (UAV) airborne platform. Background technique [0002] Low-cost small UAVs have been widely developed in all walks of life due to their advantages such as low ground support requirements, strong mobility and low safety risk factor. In the field of aerial photography, it is one of the hotspots and difficulties to perform the task of ground target detection through the camera carried by the drone. Due to the limited load of the UAV itself, it is often impossible to place a powerful GPU computing unit on the UAV's airborne platform. Currently, the commonly used UAV ground target detection method is generally offline processing, that is, the UAV only performs shooting tasks. , while the target detection part is processed on the ground computing unit in ...

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): G01C11/00G01C11/04G06N3/04G06N3/08
CPCG01C11/00G01C11/04G06N3/08G06N3/045
Inventor 胡天江李铭慧王法魁郑勋臣朱波
Owner SUN YAT SEN UNIV