Technical innovation of oil and gas pipeline safety detection based on YOLOv8 in the context of new energy

To address a series of issues such as low accuracy and low efficiency in safety detection after leakage of oil and gas transmission and storage pipelines, this paper proposes an oil and gas pipeline leakage safety detection algorithm based on YOLOv8. Firstly, by adding the SKNet attention mechanism, more feature information can be retained at a lower computational cost. Secondly, by changing the original loss function to the GWD loss function, the detection ability of the algorithm for special targets is improved, thereby realizing effective detection of oil and gas pipeline safety. Experimental results show that the improved YOLOv8 model shows significant superiority in the safety detection of oil and gas pipelines. Its accuracy reaches 88.5%, which is 6.6% higher than the original YOLOv8 model, meeting the safety detection requirements of oil and gas pipelines.