Convolution Neural Network Approach for Satellite Object Detection and Recognition

Image processing researchers have improved remote-sensing item detection. Satellite image analysis can identify houses, farms, highways, ships, airports, and hangars. An example of a more focused remote sensing imaging-based object-identification approach for military usage is a stationary aircraft detection system. The model could identify airport-parked jets. Deep learning detects airport planes in the proposed research. Google Earth satellite images dominate this collection. Deep learning modeling uses cutting-edge RCNN (Regions with Convolution Neural Network) to perform well. The training uses a custom CNN architecture throughout the system's early development. After anchoring the bounding boxes of stationary aircraft objects, RCNN detects regions. CNN training requires a large dataset of airplane images. Turkish airport satellite images validate the system. The investigation proves the model's high-performance airplane recognition effectiveness. The YOLO-v3 came in second place, offering a good compromise between accuracy and quickness despite its somewhat lower capability. The recommended aircraft identification framework uses matching bounding boxes to identify airplanes in test photographs, although the classifier network configuration utilized in the early research phase has a 98.4% success rate.