Fig 1.
Schematic diagram of the LMHD modeling framework.
Fig 2.
Repvit network structure.
Fig 3.
GSConv module structure.
Fig 4.
(a) Structure of GS bottleneck module; (b) Structure of DVoVGSCSPM module.
Fig 5.
Structure of depthwise over-parameterized convolution module.
Fig 6.
Attention mechanism without reference.
Fig 7.
Diagram of different FPN structures.
(a) FPN introduces a top-down pathway to fuse muli-scale features from level 3 to7(P3 – P7); (b) PANet adds an additional bottom- up pathway on top of FPN; (C) NAS-FPN use neural architecturesearch to find an irregular feature network topology and then repeatedly apply the same block.
Fig 8.
Adaptive convolution structure diagram.
Fig 9.
Structure of efficient localized attention mechanism.
Fig 10.
The PIoU loss of the bounding box regression.
Fig 11.
Steel surface defects.
Table 1.
Hyperparameter settings.
Table 2.
Test results comparison of different models.
Fig 12.
Comparison of the accuracy of each of the four detection models.
Table 3.
Ablation experiments.
Table 4.
Comparison of each lightweight backbone network.
Table 5.
Comparison of improvements in attention mechanisms.
Table 6.
Comparison of detection results with different loss functions.
Fig 13.
PR curve of NEU-DET on LMHD.
Fig 14.
Comparison of mAP@0.5 and loss function curves of each improved module.
Fig 15.
Comparison of the performance of each backbone network.
Fig 16.
Grad-CAM visualization before and after improvement.
(a) Initial image (b) Pre-improvement image (c) Post-improvement image.
Fig 17.
Comparison of loss functions.
Fig 18.
The prediction results of six types of defects.
(a) Ground truth. (b) RT-DETR (c) YOLOv5l (d) YOLOv8m. (e) LMHD.