Improving YOLO12 Performance Using Efficient Channel Attention For Ship Object Detection
DOI:
https://doi.org/10.30871/jaic.v10i3.13067Keywords:
Attention, Efficient Channel Attention, Object Detection, Ship, YOLO12Abstract
Ship object detection in aerial imagery remains a critical challenge due to complex marine backgrounds, varying object scales, and occlusion, which often lead to unstable model performance. This research proposes integrating the Efficient Channel Attention (ECA) module into the YOLO12-L architecture to enhance feature selectivity and prediction robustness. The model was trained for 500 epochs on the Ship Detection from Aerial Images dataset, comprising 621 images and 1,951 annotated ship instances, and performance was evaluated across five distinct random seeds to ensure statistical reliability. Quantitative results demonstrate that the proposed YOLO12-L + ECA model achieved a median Average Precision (mAP@50) of 71.32% and a Precision of 92.5%, outperforming the baseline YOLO12-L model. To evaluate statistical validity, a Paired Bootstrap Median Test with 100 resamples confirmed a statistically significant improvement in median performance (Δ = +1.01%, p = 0.02). Furthermore, the standard deviation of mAP@50 decreased from 1.1% in the baseline to 0.3% in the ECA model, representing a 72.7% reduction in performance variance. Computational efficiency analysis revealed that the ECA module introduced negligible overhead, adding merely 5 parameters (totaling 26,389,880) and keeping FLOPs constant at 89.4, while maintaining a high inference speed of 10.7 FPS (a marginal 2.5% reduction). These findings confirm that ECA effectively suppresses background noise, stabilizes detection outputs, and provides statistically significant improvements without compromising architectural efficiency. The proposed architecture offers a lightweight and reliable solution for automated maritime monitoring systems, particularly in challenging visual environments.
Downloads
References
[1] Vitor G. Santos, Diego S. Pereira, Luis B. P. Nascimento, and Pablo J. Alsina, “CNN-based Boat Detection for Environmental Protection Area Monitoring,” presented at the XXIV Congresso Brasileiro de Automática, Online, Oct. 2022. doi: 10.20906/CBA2022/3599.
[2] S. Das and Aravinth R, “Navigating the Future: Intelligent Ship Detection through Multisensor Imagery and DeepLearning,” ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci., vol. X-5/W2-2025, pp. 137–142, Dec. 2025, doi: 10.5194/isprs-annals-X-5-W2-2025-137-2025.
[3] F. Hermens, “Automatic object detection for behavioural research using YOLOv8,” Behav Res, vol. 56, no. 7, pp. 7307–7330, May 2024, doi: 10.3758/s13428-024-02420-5.
[4] X. Wang et al., “Ship feature recognition methods for deep learning in complex marine environments,” Complex Intell. Syst., vol. 8, no. 5, pp. 3881–3897, Oct. 2022, doi: 10.1007/s40747-022-00683-z.
[5] A. Galdelli, G. Narang, R. Pietrini, M. Zazzarini, A. Fiorani, and A. N. Tassetti, “Multimodal AI-enhanced ship detection for mapping fishing vessels and informing on suspicious activities,” Pattern Recognition Letters, vol. 191, pp. 15–22, May 2025, doi: 10.1016/j.patrec.2025.02.022.
[6] Y. Tan, J. Song, and C. Chu, “Detection of Small Object based on Improved-YOLOv8,” FCIS, vol. 10, no. 3, pp. 79–85, Dec. 2024, doi: 10.54097/n5rtnt71.
[7] J. He and S. Luo, “An Improved Model Based on YOLOv8 for Small Object Detection and Recognition,” Information, vol. 17, no. 2, p. 173, Feb. 2026, doi: 10.3390/info17020173.
[8] Ultralytics, “YOLO12: Attention-Centric Object Detection.” Accessed: May 10, 2026. [Online]. Available: https://docs.ultralytics.com/models/yolo12/
[9] T. Ge, B. Ning, and Y. Xie, “YOLO-AFR: An Improved YOLOv12-Based Model for Accurate and Real-Time Dangerous Driving Behavior Detection,” Applied Sciences, vol. 15, no. 11, p. 6090, May 2025, doi: 10.3390/app15116090.
[10] R. F. Puspita, M. Naufal, and F. Al Zami, “Improving YOLO Performance with Advanced Data Augmentation for Soccer Object Detection,” JAIC, vol. 9, no. 6, pp. 3601–3611, Dec. 2025, doi: 10.30871/jaic.v9i6.11256.
[11] P. Selvam, P. Shanmuga Sundari, M. Tamilselvi, T. Suresh, M. Murugappan, and M. E. H. Chowdhury, “YOLO-SAIL: Attention-Enhanced YOLOv5 With Optimized Bi-FPN for Ship Target Detection in SAR Images,” IEEE Access, vol. 13, pp. 29523–29540, 2025, doi: 10.1109/ACCESS.2025.3536621.
[12] P. C. Austin, I. Eekhout, and S. Van Buuren, “Evaluating the median p -value method for assessing the statistical significance of tests when using multiple imputation,” Journal of Applied Statistics, vol. 52, no. 6, pp. 1161–1176, Apr. 2025, doi: 10.1080/02664763.2024.2418473.
[13] F. Yuan, X. Gao, and C. Zhang, “Three-Dimensional Model Classification Based on VIT-GE and Voting Mechanism,” CMC, vol. 85, no. 3, pp. 5037–5055, 2025, doi: 10.32604/cmc.2025.067760.
[14] M. Xue, M. Chen, D. Peng, Y. Guo, and H. Chen, “One Spatio-Temporal Sharpening Attention Mechanism for Light-Weight YOLO Models Based on Sharpening Spatial Attention,” Sensors, vol. 21, no. 23, p. 7949, Nov. 2021, doi: 10.3390/s21237949.
[15] H. Ahn et al., “SAFP-YOLO: Enhanced Object Detection Speed Using Spatial Attention-Based Filter Pruning,” Applied Sciences, vol. 13, no. 20, p. 11237, Oct. 2023, doi: 10.3390/app132011237.
[16] K. Patel, C. Bhatt, and P. L. Mazzeo, “Deep Learning-Based Automatic Detection of Ships: An Experimental Study Using Satellite Images,” J. Imaging, vol. 8, no. 7, p. 182, Jun. 2022, doi: 10.3390/jimaging8070182.
[17] E. Yildirim and T. Kavzoglu, “Ship Detection in Optical Remote Sensing Images Using YOLOv4 and Tiny YOLOv4,” in Innovations in Smart Cities Applications Volume 5, vol. 393, M. Ben Ahmed, A. A. Boudhir, İ. R. Karaș, V. Jain, and S. Mellouli, Eds., in Lecture Notes in Networks and Systems, vol. 393. , Cham: Springer International Publishing, 2022, pp. 913–924. doi: 10.1007/978-3-030-94191-8_74.
[18] A. A. D. Go et al., “Comprehensive Benchmark of Yolov11n, SSD MobileNet, CenterFace, Yunet, FastMtCnn, HaarCascade, and LBP for Face Detection in Video Based Driver Drowsiness,” vol. 7, no. 3, 2025.
[19] M. N. Andrean et al., “Comparing Haar Cascade and YOLOFACE for Region of Interest Classification in Drowsiness Detection,” mib, vol. 8, no. 1, p. 272, Jan. 2024, doi: 10.30865/mib.v8i1.7167.
[20] R. D. L. Rocha and F. A. P. D. Figueiredo, “Enhancing YOLO-Based SAR Ship Detection with Attention Mechanisms,” Remote Sensing, vol. 17, no. 18, p. 3170, Sep. 2025, doi: 10.3390/rs17183170.
[21] Y. Li et al., “SOD-YOLO: Small-Object-Detection Algorithm Based on Improved YOLOv8 for UAV Images,” Remote Sensing, vol. 16, no. 16, p. 3057, Aug. 2024, doi: 10.3390/rs16163057.
[22] Z. Xu, Y. Yang, Y. Wei, and O. L. Magnagna, “Improvement of YOLOv8 for Vehicle Small Object Detection Research,” in International Conference on Artificial Intelligence, Automation and High Performance Computing, Zhuhai China: ACM, Jul. 2024, pp. 84–89. doi: 10.1145/3690931.3690946.
[23] B. Khalili and A. W. Smyth, “SOD-YOLOv8—Enhancing YOLOv8 for Small Object Detection in Aerial Imagery and Traffic Scenes,” Sensors, vol. 24, no. 19, p. 6209, Sep. 2024, doi: 10.3390/s24196209.
[24] M. Kim, J. Jeong, and S. Kim, “ECAP-YOLO: Efficient Channel Attention Pyramid YOLO for Small Object Detection in Aerial Image,” Remote Sensing, vol. 13, no. 23, p. 4851, Nov. 2021, doi: 10.3390/rs13234851.
[25] W. Luo and S. Yuan, “Enhanced YOLOv8 for small-object detection in multiscale UAV imagery: Innovations in detection accuracy and efficiency,” Digital Signal Processing, vol. 158, p. 104964, Mar. 2025, doi: 10.1016/j.dsp.2024.104964.
[26] F. García Fernández, P. De Palacios, A. García-Iruela, and L. G. Esteban, “Using Bootstrapping to Determine Artificial Neural Network Confidence Intervals—Case Study of Particleboard Internal Bond Determined from Production Data,” Applied Sciences, vol. 15, no. 8, p. 4554, Apr. 2025, doi: 10.3390/app15084554.
[27] “Ship Detection from Aerial Images.” Accessed: May 11, 2026. [Online]. Available: https://www.kaggle.com/datasets/andrewmvd/ship-detection
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Richard Christoper Subianto, Muhammad Naufal, Farrikh Alzami
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License (Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) ) that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
