Related Experiment Video
Updated: Mar 25, 2026

Automatic Laser-based Geometry Capture for Finite Element Analysis of Weld Beads
Published on: July 25, 2025
Weld defect detection based on improved YOLOv8n
Yongqi Yan1,2, Yi Liu1,2, Lingshuang Meng1,3
1State key Laboratory of Extreme Environment Optoelectronic Dynamic Testing Technology and Instrument, North University of China, Taiyuan, China.
This study introduces an enhanced YOLOv8n model for industrial weld defect detection, achieving higher accuracy and faster speeds. The improved method offers a lightweight and efficient solution for automated defect identification.
Area of Science:
- Computer Vision
- Machine Learning
- Materials Science
Background:
- Industrial weld defect detection faces challenges due to low contrast and blurred edges, impacting algorithm performance.
- Existing methods struggle to meet industrial demands for high accuracy, speed, and deployability.
Purpose of the Study:
- To develop an improved, lightweight YOLOv8n model for accurate and fast automated weld defect detection.
- To enhance feature representation and preserve subtle defect details for better detection.
Main Methods:
- Replaced the C2f module with C2f_OREPA for reduced complexity and enhanced feature extraction.
- Introduced DCDConv to preserve fine defect features and improve subtle defect detection.
- Incorporated a cross-scale feature fusion module (CCFM) for multi-scale defect detection.
Main Results:
- Achieved 87.6% mAP on a custom dataset (4.5% increase over YOLOv8n) and 82.8% mAP on the NEU-DET dataset (6.7% increase).
- Reduced model parameters by 26.9% and computational cost by 35.7%.
- Attained an inference speed of 103 FPS, outperforming mainstream object detection frameworks.
Conclusions:
- The proposed method offers an accurate, efficient, and deployable solution for industrial weld defect detection.
- Demonstrates significant practical value for automated quality control in manufacturing.
Related Concept Videos
Lumber Defects
Shakes are minor fractures that run along or across the wood's annual rings, while wane is...
Detection of Gross Error: The Q Test
Long-patch Base Excision Repair
Detection of Black Holes
Their closest cousins are neutron stars, which are composed almost entirely of neutrons packed against each other, making them extremely dense. A neutron star has the same mass as the Sun but its diameter is only a few kilometers. Therefore, the escape velocity from their surface is close to the speed of light.
Not until the 1960s, when the first neutron...
Leaky Scanning
Imperfections in Crystal Structure: Stoichiometric Point Defects

