Fig 1.
Architecture of the WaveMamba-YOLO.
Fig 2.
GLaM network structure diagram.
Fig 3.
Quad-Directional 2D Selective Scanning (SS2D) Trajectories in the PMConv Module’s State-Space Path.
Fig 4.
CHDWT network structure diagram.
Fig 5.
LWGA network structure diagram.
Fig 6.
Example Diagram of Severstal Steel Defect Types.
Fig 7.
NEU-DET Dataset Defect Types Illustration.
Fig 8.
GC10-DET Dataset Defect Types Illustration.
Fig 9.
Class Distribution of the GC10-DET Dataset.
Table 1.
Comparison Results for Three Different Necks on the Defect Dataset.
Fig 10.
Comparison results of thermogram visualization of three different necks.
Fig 11.
Radar Chart Comparison of YOLOv11n and YOLOv11n+CHDWT Across Multiple Metrics.
Table 2.
Comparative results of three different multi-scale modules on steel defect datasets.
Fig 12.
Comparison for Multi-scale Models.
Table 3.
Ablation study of waveMamba-YOLO under different lighting conditions.
Table 4.
Benchmarking detection performance across steel defect detection models.
Fig 13.
Visual Comparison of Defect Detection Performance Between YOLOv11n and WaveMamba-YOLO.
Table 5.
Performance comparison on NEU-DET and GC10-DET datasets. All models are evaluated under 640 × 640 resolution.
Fig 14.
Visual comparison of YOLOv11n and WaveMamba-YOLO on NEU-DET.