Surface Defect Detection of Electrolytic Nickel Plates Based on Improved YOLOv5

• 1、School of Intelligent Engineering and Automation,Beijing University of Posts and Telecommunications,Beijing, 100876
• 2、JINCHUAN GROUP CO., LTD.Nickel Smelting Plant Finished Product Workshop, Jinchang 737100

Abstract：In order to achieve accurate detection of surface defects on nickel plates and ensure the stability and reliability of products, this study proposes an improved YOLOv5 algorithm, Ni-YOLO, and establishes a small-scale nickel plate defect dataset for the detection of electrolytic nickel defects. The specific improvements are as follows: In the feature extraction stage, the Multi-Head Self-Attention module (MHSA) is introduced to enhance the model\'s ability to capture long-range dependencies and reduce background interference. Then, the Atrous Spatial Pyramid Pooling (ASPP) is introduced to replace the original SPPF module, which is used to enhance the model\'s receptive field and improve the detection ability for small targets and targets with high aspect ratios. The test results on the self-built dataset show that the mAP50 and mAP50-95 are 500.2% and 41.4% respectively, which are 4.5% and 2.6% higher than those of the original YOLOv5m algorithm. In addition, compared with the latest object detection algorithms YOLOv8, YOLOv9, YOLOv10, and Gold-YOLO, the improved YOLOv5 has obvious advantages in terms of performance and computational load.

Keywords： Object detection YOLOv5 Attention mechanism ASPP