Illegal Vehicle Intrusion Detection of Oil and Gas Pipelines Based on Improved YOLOv9

In view of the problem of large illegal vehicles being constructed near the oil and gas pipeline, leading to pipeline damage and oil and gas leakage, an improved YOLOv9 is proposed. First, deeply separable convolution is used to replace traditional convolution and extract features in stages to reduce network complexity and deployment cost. Secondly, Inner-SloU is used to replace the CloU loss, and the auxiliary frame with zoom factor is introduced to improve the accuracy of small target detection and enhance the effect of multi-scale vehicle classification. Finally, a global attention module is added to the front end of the network-assisted branch, combining spatial attention and channel attention to enhance the capture of the overall features of the vehicle. The objective experimental results showed that the mAP of the improved model reached 95.54%, 2.65%, 1.82% and 3.33% compared with YOLOvl0m, YOLOv8m and YOLOv5m respectively. The number of parameters was 14.7 M and the inference time was 48ms, which met the requirements of accuracy and real-time. Subjective experiments show that the improved method has lower leakage rate and false detection rate than other models, and the model can still maintain good robustness and universality.