Deep learning-based rebar detection methods, software products, and equipment

By using a deep learning-based rebar detection method, which extracts and classifies rebar features using a pre-trained model, the problem of inaccurate detection in traditional electromagnetic tomography is solved, and high-resolution rebar localization and detection is achieved.

CN121093121BActive Publication Date: 2026-06-30ZHONGBEI UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
ZHONGBEI UNIV
Filing Date
2025-08-25
Publication Date
2026-06-30

AI Technical Summary

Technical Problem

Traditional electromagnetic tomography (EMT) technology has difficulty accurately detecting the location and diameter of reinforcing bars in the health monitoring of concrete structures. It suffers from problems such as expensive equipment, complex operation, and low imaging resolution. Furthermore, the accuracy of solving the inverse problem is insufficient, leading to inaccurate reconstruction results.

Method used

A deep learning-based rebar detection method is adopted. By acquiring voltage signals and utilizing a pre-trained rebar detection model, including a rebar feature extraction network and a classification network, rebar feature data is extracted and rebar distribution images are output, thereby improving the detection accuracy.

Benefits of technology

It can accurately identify the features of steel bars in complex environments, improve the accuracy of steel bar detection, solve the problems of image artifacts and edge blurring in traditional methods, and achieve high-resolution steel bar localization.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121093121B_ABST
    Figure CN121093121B_ABST
Patent Text Reader

Abstract

This invention discloses a deep learning-based method, program product, and device for detecting reinforcing bars. The method includes: acquiring a voltage signal for detecting concrete to be tested; forming an input for a pre-trained reinforcing bar detection model based on the voltage signal, the reinforcing bar detection model including a reinforcing bar feature extraction network and a reinforcing bar classification network; extracting reinforcing bar feature data through the reinforcing bar feature extraction network based on the voltage signal; and outputting a reinforcing bar distribution image corresponding to the concrete to be tested using the reinforcing bar classification network based on the reinforcing bar feature data.
Need to check novelty before this filing date? Find Prior Art