Lightweight strip steel defect detection algorithm based on improved YOLOv7

Abstract

The precise identification of surface imperfections in steel strips is crucial for ensuring steel product quality. To address the challenges posed by the substantial model size and computational complexity in current algorithms for detecting surface defects in steel strips, this paper introduces SS-YOLO (YOLOv7 for Steel Strip), an enhanced lightweight YOLOv7 model. This method replaces the CBS module in the backbone network with a lightweight MobileNetv3 network, reducing the model size and accelerating the inference time. The D-SimSPPF module, which integrates depth separable convolution and a parameter-free attention mechanism, was specifically designed to replace the original SPPCSPC module within the YOLOv7 network, expanding the receptive field and reducing the number of network parameters. The parameter-free attention mechanism SimAM is incorporated into both the neck network and the prediction output section, enhancing the ability of the model to extract essential features of strip surface defects and improving detection accuracy. The experimental results on the NEU-DET dataset show that SS-YOLO achieves a 97% mAP50 accuracy, which is a 4.5% improvement over that of YOLOv7. Additionally, there was a 79.3% reduction in FLOPs(G) and a 20.7% decrease in params. Thus, SS-YOLO demonstrates an effective balance between detection accuracy and speed while maintaining a lightweight profile.

Keywords: D-SimSPPF; Deep learning; Lightweight network; Strip surface defect detection; YOLOv7.

Figure 1

SS-YOLO network structure.

Figure 2

The realization of depthwise separable convolution.

Figure 3

D-SimSPPF module structure.

Figure 4

SimAM network structure. SimAM can derive 3D attention weights for feature maps without introducing additional parameters. After obtaining the weights for each neuron, it directly evaluates the importance of independent neurons and assigns higher weights to important neurons, enabling the model to focus on surface defect features of steel strips and enhance detection performance.

Figure 5

NEU-DET surface defects.

Figure 6

Illustration of the enhancement effect.

Figure 7

Comparison of the YOLOv7 and SS-YOLO model training results.

Figure 8

Comparison of YOLOv7 and SS-YOLO Precision-Recall (P-R) curves.

Figure 9

Comparison of crazing defect detection. (a) Original images; (b) YOLOv5; (c) YOLOv7; (d) SS-YOLO.

Figure 10

Comparison of inclusion defect detection. (a) Original images; (b) YOLOv5; (c) YOLOv7; (d) SS-YOLO.

Figure 11

Comparison of pitted_surface defect detection. (a) Original images; (b) YOLOv5; (c) YOLOv7; (d) SS-YOLO.

Figure 12

Comparison of patches defect detection. (a) Original images; (b)YOLOv5; (c) YOLOv7; (d) SS-YOLO.

Figure 13

Comparison of rolled-in scale defect detection. (a) Original images; (b)YOLOv5; (c) YOLOv7; (d) SS-YOLO.

Figure 14

Comparison of scratches defect detection. (a) Original images; (b) YOLOv5; (c) YOLOv7; (d) SS-YOLO.