Comparative evaluation of YOLO and RT-DETR models for real-time defect detection in wood-based 3D printing

Additive manufacturing with wood-based materials offers advantages in sustainability and design versatility but faces challenges related to defect detection during the printing process. This study evaluates five object detection models (YOLOv5, YOLOv8, YOLOv10, YOLOv11, and RT-DETR) for detecting powder-bed defects in wood-based binder jetting. A dataset of 599 high-resolution images labeled with six defect categories was used to train and validate the models. On the test set, YOLOv11 achieved the highest precision (0.632), while RT-DETR obtained the highest mAP@0.5 (0.563) and recall (0.581). YOLOv8 provided a favorable trade-off between inference speed (3.2 ms per image) and detection accuracy, making it suitable for latency-sensitive monitoring. In comparison, RT-DETR achieved slightly higher accuracy but required longer inference time (17.9 ms per image), which may limit its use in strict real-time scenarios. Overall, these results highlight the trade-offs between accuracy and computational efficiency, providing guidance for deploying defect detection models in wood-based additive manufacturing.