Full
Outline
This study proposed a high-precision, high-efficiency method for surface defect detection of particleboard that is suitable for practical industrial scenarios to address the problems of high missed detection rate, slow detection speed, and insufficient ability to identify complex defects associated with traditional detection methods in current particleboard production, thereby achieving a dual optimization of detection accuracy and real-time performance.
Based on RT-DETR-r18, ablation experiments were conducted using the control variable method to verify the effectiveness of the three key modules: ECG-CFNet, FSA, and MPCA. Several mainstream lightweight backbone networks, including Swin Transformer and VanillaNet, were substituted for comparison of their training performance. The improved model was further compared with SSD, Faster-RCNN, YOLO series, and the original RT-DETR to validate its superiority. Key indicators such as mAP@50 and GFLOPs were adopted, combined with performance curves, PR curves, and visualization results, to comprehensively evaluate the detection accuracy, efficiency, and industrial applicability of the model.
Comparison experimental results on standard defect image datasets show that the proposed improved algorithm achieves 91.9% in terms of mean detection accuracy (mAP@50), which is 2.9% higher than the original RT-DETR-r18 model. At the same time, the number of parameters of the model and the computational complexity are both decreased, which suggests that while guaranteeing the improvement of the detection accuracy, the detection efficiency is optimized as well. Specifically, in the detection of typical defect types such as cracks, shavings and black spots, both show higher recognition rates and lower false alarm rates.
The improved RT-DETR algorithm significantly improves the accuracy and real-time performance of particleboard surface defect detection while maintaining low computational resource consumption, and has strong industrial deployability and application prospects.
19
5
| 科 Family | 属数 Number of genus | 种数 Number of species | 占总种数比例 Percentage of total species (%) | 属 Genus | 种数 Number of species | 占总种数比例 Percentage of total species (%) |
|---|---|---|---|---|---|---|
| 鹅膏菌科Amanitaceae | 2 | 11 | 5.26 | 鹅膏菌属 Amanita | 10 | 4.78 |
| 小菇科 Mycenaceae | 2 | 12 | 5.74 | 丝盖伞属 Inocybe | 5 | 2.39 |
| 多孔菌科 Polyporaceae | 8 | 14 | 6.70 | 蜡蘑属 Laccaria | 5 | 2.39 |
| 红菇科 Russulaceae | 3 | 23 | 11.00 | 小皮伞属 Marasmius | 6 | 2.87 |
| 小菇属 Mycena | 11 | 5.26 | ||||
| 光柄菇属 Pluteus | 5 | 2.39 | ||||
| 红菇属 Russula | 17 | 8.13 | ||||
| 栓菌属 Trametes | 5 | 2.39 |
关闭全屏
No. 86 Xueyuan South Road, Haidian District, Beijing
010-62662699(Operation and Technical Support)
qkjq@cast.org.cn
Copyright © 2025 China Association for Science and Technology. All rights reserved. For all open access content, the relevant licensing terms apply.
Sponsored by the Office of the Leading Group for Cybersecurity and Informatization of CAST, and supported by Science and Technology Review Publishing House
