An Improved Fabric Defect Detection Using Machine Learning

The textile industry’s growing demand for high-quality fabric products has brought fabric defect detection to the forefront of quality control processes. In this paper, we provide a novel method for categorizing fabric flaws that employ a Convolutional Neural Network (CNN) and deep learning capabilities. Our research showcases the exceptional performance of the CNN model, which significantly outperforms alternative methods like Faster R-CNN, SSD, and YOLOv3. By systematically employing transfer learning techniques, pre-trained models, and fine-tuning, we have fine-tuned our CNN to achieve remarkable results in accuracy, precision, and recall. The capacity of the CNN to extract complex features from fabric defect photos has the potential to transform fabric defect identification in real-world applications, ensuring product integrity and quality. This work provides a robust foundation for advancing fabric defect classification and paves the way for practical applications in the textile industry.