Published 2022-07-30
How to Cite
This work is licensed under a Creative Commons Attribution 4.0 International License.
Abstract
Angle steel is crucial in constructing the national power grid, particularly in transmission line towers. Accurate identification and recording of angle steel types are vital to prevent errors in construction processes. Traditional manual methods are prone to mistakes, necessitating an automated solution. This paper presents an improved angle steel character detection algorithm utilizing DBNet and MobileNetV3, enhanced by Coordinate Attention (CA) to replace the original attention mechanism. By collecting a specialized angle steel character dataset and implementing data augmentation and learning rate strategies, the proposed method addresses the challenges of diverse character shapes and complex backgrounds. Experimental results demonstrate that the improved algorithm significantly outperforms existing methods, achieving a remarkable increase in the F1-Score from 81.42% to 99.06%. The study validates the effectiveness of the proposed enhancements in industrial settings and lays the groundwork for further advancements in angle steel character detection.