Petroleum pipeline leakage detection method based on improved YOLOv7

The invention relates to a petroleum pipeline leakage detection method based on improved YOLOv7, and the method sequentially comprises the following steps: S1, collecting a petroleum pipeline leakage image and video, and constructing a petroleum pipeline leakage image sample set after preprocessing; s2, performing improvement based on an existing YOLOv7 neural network to form an improved YOLOv7 neural network for petroleum pipeline leakage detection; s3, inputting the petroleum pipeline leakage image sample set constructed in the step S1 into the improved YOLOv7 neural network in the step S2 for training; and S4, applying the trained model to leakage detection of the petroleum pipeline. According to the petroleum pipeline leakage detection method based on the improved YOLOv7, the requirements for high precision and high real-time performance of petroleum pipeline leakage detection can be met, environmental pollution caused by crude oil leakage can be greatly reduced, and the labor cost of routing inspection c