Lightweight unmanned aerial vehicle target detection and tracking method and device

CN120913102APending Publication Date: 2025-11-07709TH RESEARCH INSTITUTE CHINA STATE SHIPBUILDING CORP LTD
-1 Cites -1 Cited by

Patent Information

Application Number
CN202510901019.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-07-01
Publication Date
2025-11-07

AI Technical Summary

Technical Problem

Existing UAV target detection and tracking technologies face problems such as insufficient data diversity, high network complexity, target occlusion leading to track trajectory interruption, and poor performance in detecting small targets.

Method used

A lightweight UAV target detection and tracking method is adopted. By processing aerial video in frames, combining feature extraction, detection and re-identification, spatial similarity and apparent similarity are calculated, Kalman filtering is used to predict the target position, and the Hungarian algorithm is used for optimal matching. A target training set is constructed for model training to enhance data diversity and network generalization ability.

Benefits of technology

It improves the accuracy and stability of UAV target detection in complex scenarios, reduces target switching and tracking loss, and enhances autonomous perception and tracking capabilities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120913102A_ABST
    Figure CN120913102A_ABST
Patent Text Reader

Abstract

The invention relates to the field of multi-target tracking in the field of computer vision, in particular to a lightweight unmanned aerial vehicle target detection and tracking method and device. The method comprises the following steps: framing an aerial video to obtain a discrete aerial image; performing feature extraction on the aerial image to obtain a feature map; detecting the feature map to obtain a target detection result; re-identifying the feature map to obtain a target re-identification result; calculating a spatial similarity based on the target detection result; calculating an apparent similarity based on the target re-identification result; and performing multi-target data association based on the spatial similarity and the apparent similarity to obtain a tracking sequence. According to the invention, the conditions of target switching and tracking loss are reduced, and the autonomous sensing and tracking capability of the unmanned aerial vehicle in a complex scene is significantly improved.
Need to check novelty before this filing date? Find Prior Art