Enhanced sewing defect detection with YOLOv11: toward smart quality control in the apparel industry

Résumé fourni par la source

Quality control is essential in the garment industry, yet sewing defects remain less explored than fabric flaws. Manual inspection methods are still commonly used, but they are time-consuming, subjective, and error prone. This study proposes an automated method for detecting sewing defects such as broken stitches, open seams, and fabric holes using YOLOv11, a widely used object detection model in computer vision. We used the platform Roboflow for every stage of data preparation, from data preprocessing to annotation and data augmentation. The experiments were conducted on a dataset composed of 2157 images, including three sewing defect classes. The proposed approach yields promising precision results which achieve the mean Average Precision of 89.6% and an inference time of 5.7 ms per image, highlighting its potential for integration into real-time inspection systems.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.