An unmanned aerial vehicle micro-target detection method based on dynamic loss scheduling

By employing dynamic loss scheduling and numerical anti-collapse mechanisms, the problems of false detection and training instability in UAV aerial target detection were solved, achieving high-precision and stable detection results.

CN122657761APending Publication Date: 2026-08-28SOUTHWEST PETROLEUM UNIV
0 Cites 0 Cited by

Patent Information

Application Number
CN202610884231.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-06-18
Publication Date
2026-08-28

AI Technical Summary

Technical Problem

The detection of small targets in UAV aerial images suffers from problems such as missed detections, false detections, and unstable localization. Existing methods are prone to increased false alarms during high-resolution training and are unstable under limited hardware resources.

Method used

A dynamic loss scheduling-based approach is adopted, which enhances the stability and accuracy of small target detection by coordinating the design of bounding box regression loss through staged dynamic loss scheduling, logarithmic scale invariance penalty, shrinkage auxiliary box metric, and mixed precision numerical anti-collapse mechanism.

Benefits of technology

It improves the accuracy and training stability of small target detection in UAV aerial photography, reduces the false alarm rate, and is suitable for complex backgrounds and limited hardware resources.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The application discloses a kind of unmanned aerial vehicle micro target detection methods based on dynamic loss scheduling, belongs to computer vision target detection technical field, to solve the problem that the size of unmanned aerial vehicle aerial micro target is small, background is complex, shallow feature is easy to introduce false alarm, small batch mixed precision training is easy to shock and collapse.This method embeds P2 shallow high-resolution features as an implementation carrier in the target detection network, constructs a SIL logarithmic scale penalty term in the bounding box regression loss, and combines Inner-IoU, momentum smoothing focus and DLS three-stage dynamic loss scheduling;At the same time, through floating-point promotion, denominator lower bound truncation and exponential upper limit fuse, a numerical anti-collapse pipeline is formed to improve the positioning accuracy, training stability and engineering feasibility of micro targets, which can be used for unmanned aerial vehicle inspection, traffic monitoring, security patrol and low-altitude remote sensing target detection.
Need to check novelty before this filing date? Find Prior Art