The rise of state-of-the-art deep learning models and computer vision techniques have ensured the application of AI almost everywhere, similarly there are systems that are being developed for aerial surveillance for ensuring security and increasing the defense capabilities at the national front. In this work a system has been proposed on similar grounds for real time object segmentation in the videos of airplanes and helicopters, this work employs the YOLOv8 model to recognize helicopters and airplanes. Recognising the inherent flaws in old methods, deep learning offers major advantages, such as enhanced adaptability to complex situations and accuracy. The dataset, which contains 7506 training photos, and 2503 validation images, serves as the basis for training and evaluating the model. The model’s excellent performance is highlighted by evaluation criteria, particularly the mean average precision (mAP50–95), which yielded a score of 0.684. The results emphasize the model’s high precision and recall, proving its ability to recognize and classify things consistently. This research contributes to our under- standing of the YOLOv8 model’s practical applicability and high-performance capabilities in the field of aerial vehicles.