.png){kind=icon}
.png){kind=icon}
Return
Advanced machine learning based surface defect detection systems
.png){kind=icon}
OA
AI
DOI：10.1016/j.rineng.2026.110511.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Innovative data augmentation techniques: the research emphasizes novel data augmentation strategies to tackle data imbalance in machine learning models. by creating synthetic samples, the models can better learn from minority classes, enhancing their robustness and accuracy–essential for effective surface defect detection and improving overall production quality. • Advancements in machine learning: the work highlights how significant advancements in machine learning frameworks influence areas like image recognition. by leveraging deep architectures, the research improves the feature extraction process from raw data, which is pivotal for enhancing defect detection capabilities in manufacturing scenarios. • Utilization of multiple evaluative frameworks: integrating various performance metrics for thorough assessment by adapting the combined bivariate performance measure (cbpm) is one of key components for this research. by employing a comprehensive evaluation strategy, it optimizes detection algorithm performance, ensuring early defect identification, which is critical for minimizing risks associated with manufacturing processes. • Refinement of image processing techniques: the application of advanced processing techniques, like the discrete wavelet transform (dwt) and otsu’s thresholding, significantly enhances image clarity for defect detection. this synergistic approach improves the reliability of machine learning models in categorizing and assessing surface conditions, directly impacting quality control measures.
Keywords:
Surface defect detection
Machine learning
Image processing
Quality control
Artificial intelligence
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
7.9
Papers:
1.1W
Citations:
1.7W
Organization
No organization information available