Table 1.
Nematode species included in the training dataset for YOLO analysis.
| Xiphinema pachtaicum | Discolaimus | Paratylenchus nainianus |
| Acrobeloides nanus | Ditylenchus dipsaci | Pratylenchus thornei |
| Acrobeles cilliatus | Dorylaimus sp. | Psilenchus hilarulus |
| Alaimus primitivus | Filenchus thornei | Prodorylaimus sp. |
| Aphelenchus avenae | Geocenamus brevidens | Rhabditis sp. |
| Aphelenchoides sacchari | Mesocriconema xenoplax | Rotylenchus cypriensis |
| Boleodorus thylactus | Helicotylenchus multicinctus | Rotylenchulus macrosoma |
| Cephalobus persegnis | Helicotylenchus digonicus | Tylenchorhynchus cylindricus |
| Cervidellus sp. | Longidorus elongatus | Hoplolaimus galeatus |
| Clarkus papillatus | Mesodorylaimus sp. | Ditylenchus myceliophagus |
Fig. 1.
The Roboflow 3.0 (YOLOv8 architecture), YOLO-NAS, and YOLOv11 training workflow.
Fig. 2.
Training Loss Curves and Performance Metrics of YOLO-NAS.
Fig. 3.
Evaluation outputs of YOLO-NAS (A): Confusion matrix of class predictions. (B) F1 score distribution across test images.
Fig. 4.
Training loss curves and performance metrics of YOLOv11.
Fig. 5.
Evaluation outputs of YOLOv11 (A): Confusion matrix of class predictions. (B) F1 score distribution across test images.
Fig. 6.
Training loss curves and performance metrics of Roboflow 3.0.
Fig. 7.
Evaluation outputs of Roboflow 3.0 (A): Confusion matrix of class predictions. (B) F1 score distribution across test images.