Research on the Detection Method of Crude Oil Leaking on Offshore Platform based on the Improved YOLOv5s Model

Abstract

Aiming at the demand of crude oil leakage monitoring on offshore platforms, model optimisation and crude oil leakage target detection are studied based on the YOLOv5s model. The SE attention mechanism is introduced into the backbone network of the YOLOv5s model to enhance the model's learning of small crude oil leakage targets and suppress the background interference information. The deformable convolution module DCNv2 is introduced for controlling the sampling position of the convolution kernel to reduce the influence of irrelevant factors and further improve the detection accuracy of the model for small targets. The CIOU loss function of the original YOLOv5s is replaced with the EIOU loss function, which makes the model pay more attention to the difficult-to-detect targets and improves the detection accuracy of the model for crude oil spill targets. The longitudinal comparison experiments before and after the optimisation of the YOLOv5s model, as well as the cross-sectional comparison experiments between the optimised model and other typical models, confirm the superiority of the adopted optimisation method.