LRSAA: Large-Scale Remote Sensing Image Target Recognition and Automatic Annotation

This paper introduces a novel method for object recognition and automatic labeling in large-area remote sensing images, called LRSAA. The proposed method integrates the YOLOv11 and MobileNetV3-SSD object detection algorithms through ensemble learning to enhance overall model performance. Additionally, it utilizes Poisson disk sampling segmentation techniques along with the EIOU metric to optimize both the training and inference processes of segmented images, culminating in an integrated results framework. This approach not only minimizes computational resource requirements but also achieves an effective balance between accuracy and processing speed. The source code for this project is publicly accessible at https://github.com/anaerovane/LRSAA.