Related Experiment Video
Updated: Sep 9, 2025
05:40
Fabricating Cotton Analytical Devices
Published on: August 30, 2016
6.8K
Research on real-time detection of fabric defects based on an improved Elo rating algorithm
1Zhejiang Shuren University, Hang Zhou, Zhejiang, 310015, China. yxb71520@163.com.
Scientific Reports
|September 1, 2025
Summary
An improved Elo rating algorithm enhances fabric defect detection efficiency for real-time industrial needs. This refined algorithm achieves over 80% accuracy, outperforming other models in speed and precision.
Area of Science:
- Materials Science
- Computer Vision
- Artificial Intelligence
Background:
- Complex textured fabric defects pose challenges for traditional detection methods.
- Real-time detection is crucial for industrial manufacturing efficiency.
Purpose of the Study:
- To improve the detection efficiency of complex textured fabric defects.
- To meet the demands of real-time industrial detection.
Main Methods:
- Refinement of the traditional Elo rating algorithm, focusing on threshold calculation and detection.
- Incorporation of integral images to reduce computational load and increase speed.
- Application of the algorithm to three fabric sample types for defect detection.
Main Results:
- Achieved an overall detection accuracy exceeding 80% for fabric defects.
- Parameter analysis identified optimal sub-region counts (R) between 10 and 30.
- Outperformed five other object detection models in mean Average Precision (mAP) and Frames Per Second (FPS), achieving 102.1 FPS.
Conclusions:
- The improved Elo rating algorithm effectively detects complex textured fabric defects.
- The algorithm meets industrial requirements for real-time detection with high accuracy and speed.
