Research on Fabric Defect Detection Algorithm Based on YOLO-V8

Modern fabric production has basically achieved automation, yet there remains a significant gap in high-efficiency and precise intelligent defect detection equipment. To address the above-mentioned issues, this paper proposes a fabric defect detection algorithm based on the YOLO-V8 network. Firstly, the fabric image data is collected and standardized. Data cleaning, denoising, outlier removal are carried out, and the data is classified according to different defect types. Subsequently, a variety of data augmentation methods (such as flipping, interpolation, brightness adjustment, noise reduction, and sharpening mask) are adopted to enrich the data samples and enhance the generalization ability of the model. In terms of model construction, a neural network based on YOLO-V8 is constructed in this paper to achieve efficient real-time detection. To better optimize the model classification effect, the cross-entropy loss function is introduced to accurately measure the error between the predicted value and the actual label. In addition, the Adam optimizer is used to ensure the rapid convergence of the model during the training process and reduce the risk of over-fitting. After the model training, the performance of the model is evaluated through the confusion matrix and classification report to detect the classification effect of each defect category. The experimental results show that the recognition accuracy of the proposed method in fabric defect detection can reach over 90% for good products, holes, and oil stains, while the detection accuracy for foreign objects and thread errors is slightly lower, ranging from 87% to 88%. Compared with traditional models, the rich data distribution and targeted enhancement operations not only improve the diversity of the data but also significantly enhance the overall detection accuracy of the model.