Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Fig 1.

Schematic diagram of the LMHD modeling framework.

More »

Fig 1 Expand

Fig 2.

Repvit network structure.

More »

Fig 2 Expand

Fig 3.

GSConv module structure.

More »

Fig 3 Expand

Fig 4.

DVoVGSCSPM network structure.

(a) Structure of GS bottleneck module; (b) Structure of DVoVGSCSPM module.

More »

Fig 4 Expand

Fig 5.

Structure of depthwise over-parameterized convolution module.

More »

Fig 5 Expand

Fig 6.

Attention mechanism without reference.

More »

Fig 6 Expand

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.

More »

Fig 7 Expand

Fig 8.

Adaptive convolution structure diagram.

More »

Fig 8 Expand

Fig 9.

Structure of efficient localized attention mechanism.

More »

Fig 9 Expand

Fig 10.

The PIoU loss of the bounding box regression.

More »

Fig 10 Expand

Fig 11.

Steel surface defects.

More »

Fig 11 Expand

Table 1.

Hyperparameter settings.

More »

Table 1 Expand

Table 2.

Test results comparison of different models.

More »

Table 2 Expand

Fig 12.

Comparison of the accuracy of each of the four detection models.

More »

Fig 12 Expand

Table 3.

Ablation experiments.

More »

Table 3 Expand

Table 4.

Comparison of each lightweight backbone network.

More »

Table 4 Expand

Table 5.

Comparison of improvements in attention mechanisms.

More »

Table 5 Expand

Table 6.

Comparison of detection results with different loss functions.

More »

Table 6 Expand

Fig 13.

PR curve of NEU-DET on LMHD.

More »

Fig 13 Expand

Fig 14.

Comparison of mAP@0.5 and loss function curves of each improved module.

More »

Fig 14 Expand

Fig 15.

Comparison of the performance of each backbone network.

More »

Fig 15 Expand

Fig 16.

Grad-CAM visualization before and after improvement.

(a) Initial image (b) Pre-improvement image (c) Post-improvement image.

More »

Fig 16 Expand

Fig 17.

Comparison of loss functions.

More »

Fig 17 Expand

Fig 18.

The prediction results of six types of defects.

(a) Ground truth. (b) RT-DETR (c) YOLOv5l (d) YOLOv8m. (e) LMHD.

More »

Fig 18 Expand