Generator rotor vent hole foreign matter monitoring and identifying system and method

The invention discloses a rotor vent hole foreign matter identification method and system based on lightweight YOLOv4, and the method comprises the steps: collecting a vent hole foreign matter image, and carrying out the preprocessing of the image; constructing a ventilation hole foreign matter data set; constructing a lightweight foreign matter recognition model; based on the ventilation hole foreign matter data set, a lightweight foreign matter recognition model is trained; and testing the trained lightweight foreign matter identification model, and identifying the foreign matter in the vent hole. According to the method, various illumination brightness environments are simulated and corresponding video pictures are collected, so that the model learns rules implied behind the images, and data except a learning set with the same rule can also be properly output through a trained network; when the model faces unknown data, the model also has good prediction capability, and over-fitting and under-fitting are e