Leveraging Deep Learning for Accurate Aircraft Classification in Remote Sensing Images
Halise Nur Aydin, Begüm ERKAL, Çağatay Berke Erdaş · 2026
The classification of aircraft types from satellite imagery is of great importance in several fields. It plays a vital role in aircraft tracking and management, and in security control at airports and military bases. It can also be used to detect different types of aircraft in hostile territory or border areas, determine military strategies, and analyze potential threats. In cases of natural disasters, identifying aircraft types can help to ensure that relief, search, and rescue operations are carried out effectively. In essence, the classification of aircraft types can provide critical information and aid strategic decision-making in many areas. The development of such technologies and their support with artificial intelligence-based models in data analytics can contribute to obtaining more accurate and faster results. In this study, deep learning models were used to classify aircraft types. These models were trained and tested on the MTARSI dataset. MTARSI is a dataset that has images of 20 aircraft types and consists of 9,589 aircraft images in total. Looking at the evaluation results obtained with the RegNetX032model, an accuracy value of 0.9146 was obtained.