Aircraft segmentation algorithm based on Unet and improved Yolov4

In order to overcome the problems of poor effect and low accuracy of semantic segmentation of aircraft target, a small target segmentation method based on improved Yolov4 and Unet is proposed. This method combines Yolov4, Unet and Mobilenet. In order to reduce the parameters of the model, Mobilenet is used to improve the Yolov4 backbone network. The improved Yolov4 is used to extract the ROI in the image, then the image is sent to VGG16-Unet for segmentation, finally, the segmented sub objects are fused with the original image to obtain a complete segmentation image. The RSOD-Dataset is used as the training dataset, and UCAS-AOD dataset is used to verify the generalization ability of the model. The experimental results show that the model has a good segmentation effect on the aircraft target in the image.