Article Details

Abstract

This paper addresses the challenges of low efficiency, high missed detection rate, and inconsistent detection standards in manual defect detection during white cloth production in the textile industry. An embedded white cloth defect detection system based on the RT-Thread operating system is designed. The system innovatively integrates a two-level visual detection strategy of “coarse detection–fine detection” in its architecture: the front-end OpenMV module completes high-speed defect coarse detection and triggering, and the back-end STM32H743 main controller, based on the RT-Thread NNOM software package, achieves the deployment of a multi-class white cloth defect end-side INT8 quantization model and real-time fine detection classification. At the same time, the system integrates stepper motor transmission alignment, automatic marking, local Excel storage, and OneNET cloud platform upload functions, forming an integrated closed-loop detection solution. Test results demonstrate that the system achieves rapid detection and accurate classification of various defects such as cracks, oil stains, and glue, with a coarse detection response time of ≤180 ms and a fine detection classification accuracy of ≥94%, meeting the real-time detection requirements of textile production lines. Compared with traditional manual detection, it significantly improves efficiency and accuracy, providing a low-cost, lightweight, and highly integrated intelligent detection solution for small and medium-sized textile enterprises.

Keywords

RT-Thread; STM32H743VGT6; defect detection; embedded machine vision; Deep Learning; OneNET cloud platform

References

1. [1] Zhao Yanting Research on Textile Surface Defect Detection Algorithm Based on Machine Vision [D]. Wuhan: Wuhan Textile University, 2025
2. [2] Zhao Xianli, Wang Zhiming, Tian Pengfei Design of Corn Seedling Grading Detection System Based on Embedded Machine Vision [J]. Agricultural Mechanization Research, 2023, 45 (9): 66-70
3. [3] Sun Yuan, Liu Chong, Ji Xuezhi, etc Design of a blood dialysis machine calibration device based on RT Thread [J]. China Testing, 2023, 49 (S2): 108-112
4. [4] Huang Y B，Xiang Z. RPDNet: Automatic Fabric Defect Detection Based on a Convolutional Neural Network and Repeated Pattern Analysis [J/OL].Sensors，2022，22（16）：6226.https://doi.org/10.3390/s22166226.
5. [5] Wang Yapei, Wei Zhe, Zhang Kai Lightweight detection algorithm for bearing surface defects based on improved YOLOv11 [J/OL]. Computer Application, 2026:1-14 https://link.cnki.net/urlid/51.1307.TP.20260416.0917.002.
6. [6] Wang Biao, Gao Yuanchen, Fang Sichao, etc Design of TDLAS CH4 gas detection data acquisition system based on STM32H743 [J]. Laser Journal, 2024, 45 (12): 45-48
7. [7] Xu Wuji, Wang Zheng, Zou Jiamin, etc Design and Implementation of Motion Target Control and Automatic Tracking System Based on OpenMV Camera [J]. Modern Electronic Technology, 2024, 47 (17): 166-172
8. [8] Xu Lihua, Wang Yihuai, Xi Shengxin Analysis of RT Thread Context Switching Mechanism Based on Cortex-M4 Core [J]. Computer Applications and Software, 2025, 42 (1): 213-216302
9. [9] Cheng Lan Design of General Control Mini Program for Smart Home Model Based on OneNET IoT Cloud Platform and MQTT Protocol [J]. Computer and Telecommunications, 2022 (7): 10-13
10. [10] Li Meng, He Hongfei Design and Implementation of Task Scheduling System for OneOS [J]. Microcontrollers and Embedded System Applications, 2022, 22 (9): 13-17, 21
11. [11] Sun Lifeng Comparative Analysis of Task Switching Methods in Embedded Operating Systems [J]. Microcontrollers and Embedded System Applications, 2011, 11 (7): 10-12, 16

Funding

2025 Hebei Province Provincial College Students Innovation and Entrepreneurship Training Program Funding Project (S202510081042)

About the authors

Tang Zhe (2005-), male, Han ethnicity, from Zhangjiakou, Hebei Province, is currently pursuing a bachelor's degree. His research interests include visual detection and optimization algorithms; Wei Longhui (2005-), male, Han ethnicity, from Xinyang, Henan Province, undergraduate studies, research direction: mechanical structure optimization; Jia Qinglin (2005-), male, Han ethnicity, from Cangzhou, Hebei Province, undergraduate studies, research direction: model embedded deployment; Liu Tianqi (2005-), female, Han ethnicity, from Daqing, Heilongjiang Province, undergraduate studies, research direction: program optimization; Corresponding author: Chen Lei (1981-), male, Han ethnicity, from Nanyang, Henan Province, associate professor, master's student, research direction: detection technology and intelligent devices, artificial intelligence technology, new energy generation and smart grid.