Quality control is an area of utmost importance for fabric production companies. By not detecting the defects present in the fabrics, companies are at risk of losing money and reputation with a damaged product. In order to reduce these costs, an automatic textile defect detection embedded system is proposed. To perform the task of defect detection, a Convolutional Neural Network: Mask RCNN and one of the latest YOLO algorithms: YOLOv8 were used in this work. To obtain an embedded system capable of detecting static and moving defects, two datasets of static images and videos were used. In our experiments, we used the best textile defect detection training on static images, from the two pre-trained models in order to detect textile defects in videos. A confidence of 66%–97% was achieved with the two-stage model Mask RCNN and a confidence of 44%–80% was achieved with the one-stage model YOLOv8 in detecting defects within videos. But YOLOv8 was significantly faster than Mask RCNN in terms of speed detection per video.