Directed target detection method based on hierarchical masking and robust ensemble regression

CN118470398BActive Publication Date: 2026-05-26NORTHWESTERN POLYTECHNICAL UNIV +1
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
NORTHWESTERN POLYTECHNICAL UNIV
Filing Date
2024-05-10
Publication Date
2026-05-26

AI Technical Summary

Technical Problem

In remote sensing image target detection, complex backgrounds and arbitrary target orientations lead to inaccurate target feature extraction and increased regression complexity. Existing methods cannot effectively improve detection performance, especially in oriented target detection where there is a lack of perception of angle changes.

Method used

A directed object detection method based on hierarchical masking and robust ensemble regression is adopted. By constructing a directed object detection network and introducing a hierarchical masking module and robust ensemble regression constraints, the accuracy of feature extraction and bounding box regression is improved.

Benefits of technology

It effectively suppresses background noise, improves detection performance and efficiency, and is suitable for target detection tasks in aerial images and scene text.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118470398B_ABST
    Figure CN118470398B_ABST
Patent Text Reader

Abstract

This invention provides a directed object detection method based on hierarchical masking and robust ensemble regression. First, images from a publicly available directed object detection dataset are cropped to obtain image patches of the same size, which are then used as training images for the network. Next, a directed object detection network is constructed, using the image as input and the predicted bounding box of the image as output, and the network is trained. Finally, the image to be detected is input into the trained directed object detection network to obtain the object detection result. This invention effectively improves upon existing directed object detection methods, addressing their poor background noise reduction and lack of perception of directed object angles, thereby enhancing detection performance and achieving good detection efficiency.
Need to check novelty before this filing date? Find Prior Art