Motion Tracking Detection and Tracking Technology Based on Aerial Video

Aiming at the problems of complex outdoor background lighting conditions, unsatisfactory moving target features, and high degree of background fusion of wind turbines with large target background, the proposed wind turbine detection and tracking method based on deep learning convolutional neural network is proposed. This paper uses migration learning technology to use a pre-trained SSD model trained with COCO data sets and makes a dedicated localized wind turbine dataset . Through data enhancement, the neural network model can effectively improve the accuracy of wind turbine identification and the average accuracy remains above 96%. The fast recognition speed enables detection and tracking of the fan target in aerial video quickly and accurately.