Write your message
                  Back to the articles list | Back to browse issues page

XML Print

A Flexible Hybrid Attention Mechanism for Multi-Architecture Segmentation of Small Maritime Targets in South East Region of Iran
Zobeir Raisi *1 , Esmaeil Sarani1 , Rasoul Damani1 , Valimohammad Nazarzehi Had1
1- Chabahar Maritime University, Chabahar, Iran
Abstract:   (118 Views)
Maritime vessel detection in satellite imagery is essential for coastal monitoring, traffic regulation, and maritime security. Vessels along the southeastern coast of Iran exhibit unique structural and geometric characteristics; they are small and overlapped, differing substantially from international benchmarks. Detecting small and overlapping vessels presents additional challenges due to the loss of fine-grained features and ambiguous object boundaries in conventional deep learning pipelines. Consequently, existing pre-trained models, trained primarily on global datasets, often fail to generalize effectively to this region. To address this, in our study, we provide the first systematic investigation of ship detection for southeastern Iran, supported by a curated dataset of high-resolution satellite imagery from its major ports. We, then, propose a flexible Hybrid Attention Fusion (HAF) module that can be seamlessly integrated into multiple segmentation architectures, including FPN, Mask R-CNN, U-Net, and DeepLab. The module sequentially applies channel and spatial attention mechanisms to adaptively recalibrate multi-scale features, enhancing the representation of subtle and occluded instances. Experimental results demonstrate that HAF-augmented models significantly outperform their baseline counterparts across all architectures. For semantic segmentation, U-Net+HAF and DeepLabv3+HAF achieve mean IoU improvements of 4.5% and 4.4%, respectively, reaching 83.8% and 85.5% mIoU. For instance segmentation, Mask R-CNN+HAF demonstrates the most substantial improvement in small object detection, with Average Precision for small objects (APs) increasing from 42.3% to 50.6%—an 8.3-point improvement. Qualitative analysis confirms superior capability in detecting missed small instances, separating overlapping vessels, and producing more precise boundaries compared to baseline models.
Keywords: Satellite images, ship detection and segmentation, deep learning, South East Iran, Attention Mechanism
Full-Text [PDF 633 kb]   (42 Downloads)    

Highlights
  • Introduces a Hybrid Attention Fusion (HAF) module combining channel and spatial attention for improved small‑object segmentation. 
  • Demonstrates 4–9% accuracy gains across multiple architectures (U‑Net, DeepLabv3, FPN, Mask R‑CNN). 
  • Achieves 67.9% overall AP and 50.6% APs for small maritime targets. 
  • Provides a flexible, architecture‑agnostic module adaptable to various segmentation models. 
  • Employs a hybrid loss (λ = 0.6) to optimize detection of small, overlapping objects in satellite imagery.

Type of Study: Research Paper | Subject: Maritime Transport and Port Management
Received: 2025/09/22 | Accepted: 2026/03/28
Audio file of the article abstract [MP3 4297 KB]  (8 Download)
References
1. T. Zhao et al.,(2024), "Ship Detection with Deep Learning in Optical Remote-Sensing Images: A Survey of Challenges and Advances," Remote Sens., vol. 16, no. 7, p. 1145, Mar. 2024, doi: 10.3390/rs16071145. [DOI:10.3390/rs16071145]
2. Z. Raisi et al.,(2026), "MVSegNet: A Multi-Scale Attention-Based Segmentation Algorithm for Small and Overlapping Maritime Vessels," Algorithms, 19(1), 23. 2026, doi: [DOI:10.3390/a19010023]
3. M. Bakirci,(2024), "Advanced ship detection and ocean monitoring with satellite imagery and deep learning for marine science applications," Reg. Stud. Mar. Sci., vol. 81, p. 103975, doi: 10.1016/j.rsma.2024.103975. [DOI:10.1016/j.rsma.2024.103975]
4. C. Zhang et al.,(2024), "Development and Application of Ship Detection and Classification Datasets: A review," IEEE Geosci. Remote Sens. Mag., vol. 12, no. 4, pp. 12-45, doi: 10.1109/MGRS.2024.3450681. [DOI:10.1109/MGRS.2024.3450681]
5. A. Mazzeo, A. Renga, and M. D. Graziano,(2024), "A Systematic Review of Ship Wake Detection Methods in Satellite Imagery," Remote Sens., vol. 16, no. 20, p. 3775, doi: 10.3390/rs16203775. [DOI:10.3390/rs16203775]
6. L. Bui et al.,(2024), "UOW-Vessel: A Benchmark Dataset of High-Resolution Optical Satellite Images for Vessel Detection and Segmentation," in 2024 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), Waikoloa, HI, USA: IEEE, pp.4416-4424,doi:10.1109/WACV57701.2024.00437. [DOI:10.1109/WACV57701.2024.00437]
7. T.-Y. Lin, P. Dollár, R. Girshick, K. He, B. Hariharan, and S. Belongie, (2017), "Feature Pyramid Networks for Object Detection," in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 2117-2125. doi: 10.1109/CVPR.2017.106. [DOI:10.1109/CVPR.2017.106] []
8. K. He, G. Gkioxari, P. Dollár, and R. Girshick, (2017), "Mask R-CNN," in IEEE International Conference on Computer Vision (ICCV), pp. 2980-2988. [DOI:10.1109/ICCV.2017.322]
9. O. Ronneberger, P. Fischer, and T. Brox, (2015), "U-Net: Convolutional Networks for Biomedical Image Segmentation," in Medical Image Computing and Computer-Assisted Intervention - MICCAI, vol. 9351. [DOI:10.1007/978-3-319-24574-4_28]
10. L.-C. Chen, G. Papandreou, I. Kokkinos, K. Murphy, and A. L. Yuille, (2018), "DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs," IEEE Trans. Pattern Anal. Mach. Intell., vol. 40, no. 4, pp. 834-848, doi: 10.1109/TPAMI.2017.2699184. [DOI:10.1109/TPAMI.2017.2699184] [PMID]
11. D. Wei, Q. Shi, and X. Li, (2020), "Ship Detection in Optical Remote Sensing Images: A Review," IEEE Geosci. Remote Sens. Mag., vol. 8, no. 4, pp. 37-52, doi: 10.1109/MGRS.2020.2999236.
12. C. Corbane, L. Najman, E. Pecoul, L. Demagistri, and M. Petit, (2010), "A complete processing chain for ship detection using optical satellite imagery," Int. J. Remote Sens., vol. 31, no. 22, pp. 5837-5854, doi: 10.1080/01431161.2010.512310. [DOI:10.1080/01431161.2010.512310]
13. F. Bovolo, C. Bruzzone, and L. Bruzzone, (2008), "A Novel Approach to Unsupervised Change Detection Based on a Semisupervised SVM and a Similarity Measure," IEEE Trans. Geosci. Remote Sens., vol. 46, no.7, pp.2070-2082, doi: 10.1109/TGRS.2008.916644. [DOI:10.1109/TGRS.2008.916644]
14. Z. Zou and Z. Shi, (2016), "Ship Detection in Spaceborne Optical Image With SVD Networks," IEEE Trans. Geosci. Remote Sens., vol. 54, no. 10, pp. 5832-5845, doi: 10.1109/TGRS.2016.2572736. [DOI:10.1109/TGRS.2016.2572736]
15. S. Wang, Y. Wang, J. Li, G. Zhao, and Z. Zhang, (2019),"A Robust Ship Detection Algorithm via Convolutional Neural Networks for SAR Images," Int. J. Remote Sens., vol. 40, no. 12, pp. 4665-4680, doi: 10.1080/01431161.2019.1570391.
16. Y. Liu, Y. Li, Y. Yuan, and Z. Wang, (2017), "HRSC2016: A High Resolution Ship Collection Dataset for Object Detection in Remote Sensing Images," in Proceedings of the International Conference on Pattern Recognition (ICPR), pp. 2310-2315. doi: 10.1109/ICPR.2017.800.
17. Airbus Defence and Space, (2018), "Airbus Ship Detection Challenge.". [Online]. Available: https://www.kaggle.com/c/airbus-ship-detection
18. S. Ren, K. He, R. Girshick, and J. Sun, (2017), "Faster R-CNN: Towards real-time object detection with region proposal networks," IEEE Trans. Pattern Anal. Mach. Intell., vol. 39, no. 6, pp. 1137-1149. [DOI:10.1109/TPAMI.2016.2577031] [PMID]
19. J. Redmon and A. Farhadi, (2018), "YOLOv3: An Incremental Improvement," ArXiv Prepr. ArXiv180402767.
20. W. Liu et al., (2016),"SSD: Single Shot MultiBox Detector," in Proceedings of the European Conference on Computer Vision (ECCV), in LNCS, vol. 9905, pp. 21-37. doi: 10.1007/978-3-319-46448-0_2. [DOI:10.1007/978-3-319-46448-0_2]
21. X. Xu, Y. Zhang, and Z. Li, (2020), "Small Ship Detection in High-Resolution Satellite Images Using Deep Learning," Remote Sens., vol. 12, no. 12, pp. 1993-2008, doi: 10.3390/rs12121993. [DOI:10.3390/rs12121993]
22. A. Vaswani et al., (2017), "Attention Is All You Need," in Advances in Neural Information Processing Systems (NeurIPS), pp. 5998-6008.
23. J. Ba, V. Mnih, and K. Kavukcuoglu, (2015), "Multiple Object Recognition with Visual Attention," in Proceedings of the International Conference on Learning Representations (ICLR).
24. J. Hu, L. Shen, and G. Sun, (2018),"Squeeze-and-Excitation Networks," in 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT: IEEE, pp. 7132-7141. doi: 10.1109/CVPR.2018.00745. [DOI:10.1109/CVPR.2018.00745]
25. S. Woo, J. Park, J.-Y. Lee, and I. S. Kweon, (2018), "CBAM: Convolutional Block Attention Module," arXiv. doi: 10.48550/ARXIV.1807.06521.
26. X. Li, Z. Li, H. Wang, and L. Jiao, (2018), "Attention-guided convolutional neural network for ship detection in SAR images," Remote Sens., vol. 10, no. 3, pp. 1-17.
27. L. Huyan et al., (2021), "A Lightweight Object Detection Framework for Remote Sensing Images," Remote Sens., vol. 13, no. 4, p. 683, doi: 10.3390/rs13040683. [DOI:10.3390/rs13040683]
28. J. Wang, J. Ding, H. Guo, W. Cheng, T. Pan, and W. Yang, (2019),"Mask OBB: A semantic attention-based mask-oriented bounding box representation for multi-category object detection in aerial images," Remote Sens., vol. 11, no. 24, p. 2930. [DOI:10.3390/rs11242930]
29. C. H. Sudre, W. Li, T. Vercauteren, S. Ourselin, and M. Jorge Cardoso, (2017),"Generalised Dice Overlap as a Deep Learning Loss Function for Highly Unbalanced Segmentations," in Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support, vol. 10553 [DOI:10.1007/978-3-319-67558-9_28] [PMID] []
30. T.-Y. Lin, P. Goyal, R. Girshick, K. He, and P. Dollar, (2020), "Focal Loss for Dense Object Detection," IEEE Trans. Pattern Anal. Mach. Intell., vol. 42, no. 2, pp. 318-327, doi: 10.1109/TPAMI.2018.2858826. [DOI:10.1109/TPAMI.2018.2858826] [PMID]
31. Q. Zhu, (2023), GeoAI: Artificial Intelligence for Geospatial Data. GitHub; 2023. [Online]. Available: https://github.com/opengeos/geoai.
Send email to the article author

Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Creative Commons License
International Journal of Maritime Technology is licensed under a

Creative Commons Attribution-NonCommercial 4.0 International License.