14 February 2024 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
With the advancements in Space technology and the development of light-weight Synthetic Aperture Radar (SAR) satellites by commercial companies such as ICEYE, Capella Space and Umbra, SAR images have become available on a wide scale. Ship detection is a classic problem in the interpretation and analysis of satellite images and has its significance both in maritime as well as defense applications. In the case of SAR images, ship detection becomes even more challenging due to the presence of large-scale distortions as well as inter-class similarity signature problem. Moreover, the State-of-the-Art (SOTA) object detection models have weak generalization capability over SAR datasets. To overcome these challenges, we propose a You Only Look Once (YOLO) based, optimized ship detection model called YOLO-OSD. Our optimized ship detector is based on a hybrid data-model centric approach which utilizes the statistical characteristics of the datasets under observation and has an efficient model architecture. We also carry out a detailed comparative analysis of our proposed model with other SOTA deep learning models on three well-known publicly available datasets. Our results show that the proposed YOLO-OSD outperforms YOLO5, YOLO7 and RetinaNet on all datasets under observation in terms of F1 score and mean Average Precision (mAP). YOLO-OSD also has approximately 16% fewer network parameters as compared to the original YOLO5. Moreover, our proposed model is at least 37.7% faster than YOLO7 and 41.02% faster than the YOLO8 model in terms of training time and thus suitable for real-time satellite based SAR ship detection.
27 March 2025 in IEEE Aerospace and Electronic Systems Magazine
Civilian and military aircraft are increasingly vulnerable to passive missile threats, such as short-range or within-visual-range air-to-air missiles (SRAAMs/WVRAAMs) and man-portable air defense systems (MANPADS). These missiles evade detection by an aircraft's radar warning receiver (RWR) due to their lack of radio frequency (RF) emissions. This paper presents a novel deep learning-based passive missile detection algorithm using simulated solar blind ultraviolet (SBUV) signatures, which offer unique advantages over traditional infrared (IR) signatures. The algorithm is built on an improved YOLOv5 (You Only Look Once) framework, capable of detecting and classifying UV signatures in real-time from sequential image data. The architecture of YOLOv5 has been modified for improved SBUV detection by recalculation of anchor box sizes, non-maximum suppression (NMS) threshold re-adjustment, SBUV specific data augmentations and increased resolution in detection heads. To overcome the challenge of limited training data, we employ advanced data synthesis techniques to create realistic training datasets derived from 3D missile and aircraft combat scenario simulations in the SBUV spectrum. These simulations incorporate diverse parameters and conditions to closely replicate real-world scenarios, ensuring high fidelity and robustness. Performance evaluation against real-world scenarios revealed an F1-score of 95% and a mean average precision (mAP) of 95% for synthetic data, compared to 88.36% and 85%, respectively, for real-world data. The algorithm, achieved a detection accuracy of 92% on synthetic data and 86% on real-world data, demonstrating its reliability and robustness. Validation under varying environmental conditions demonstrated optimal performance under clear skies, achieving benchmark results. Minimal performance degradation was observed under low-light conditions, such as dusk or dawn, as well as fog. However, overcast conditions posed the most significant challenge, reducing detection accuracy from 86% to 79% and the F1-score from 88.36% to 80%. The processing speed of the proposed algorithm satisfies the real-time requirements of modern missile warning systems, making it a potential candidate for operational deployment.
21 April 2025 in Multimedia Tools and Applications
13 October 2026 in Expert Systems