YOLOShipTracker: Tracking ships in SAR images using lightweight YOLOv8
Bibliographic record
Abstract
• A YOLOShipTracker model was developed for ship tracking using the YOLOv8 model. • HGNetv2 Backbone and Lightweight Neck improve SAR feature extraction. • Decoupled head and knowledge distillation reduce parameters and maintain accuracy. • Development of C-BIoU for accurate tracking with high performance and real-time capability. This paper presents a novel approach to tracking ships in Synthetic Aperture Radar (SAR) images based on an improved lightweight YOLOv8 Nano (YOLOv8n), specially devised to improve efficiency without compromising accuracy. In our method, we replaced the heavy backbone and neck of YOLOv8 with HGNetv2 and slim-neck, respectively. We also implemented a lightweight decoupling head using EMSConvP. Additionally, we integrated a knowledge distillation module to further enhance detection capabilities. Furthermore, we conducted extensive experiments on the short-time sequence SAR dataset to demonstrate superior accuracy metrics compared to the original YOLOv8n model. Regarding tracking ships in SAR images, we developed a multi-object tracking (MOT) technique called Cascaded-Buffered IoU (C-BIoU). This method enlarges the detection and trajectory matching space by increasing the buffer zone, effectively combining detection and trajectory information from short-time sequence SAR images. The findings reveal that our method significantly reduces the computational complexity, parameters, and model size by up to 54.7 %, 68.4 %, and 68.3 %, respectively, with respect to the original model metrics. As a direct consequence of these reductions, our proposed model demonstrates a remarkable 133.1 % improvement in image processing speed expressed as frames per second (FPS). Moreover, Our C-BIoU method shows outstanding performance in tracking accuracy and efficiency, with superior Higher Order Tracking Accuracy (HOTA), Multiple Object Tracking Precision (MOTP), and Identification F1 score (IDF1) scores of 72.8 %, 87.9 %, and 80.7 %, respectively, compared to existing tracking algorithms. The results from testing on multiple datasets highlight our method’s excellent performance in ship detection and tracking, offering high-speed processing capabilities with an average image processing speed of 81 FPS. In this sense, this method provides reliable real-time monitoring and management of maritime traffic, enhancing situational awareness for maritime operations.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".