MarineYOLO: Innovative deep learning method for small target detection in underwater environments DOI Creative Commons

Linlin Liu,

Chengxi Chu,

Chuangchuang Chen

et al.

Alexandria Engineering Journal, Journal Year: 2024, Volume and Issue: 104, P. 423 - 433

Published: Aug. 7, 2024

In the realm of underwater object detection, conventional methodologies often encounter challenges in accurately identifying and detecting small targets. These difficulties stem primarily from intricate nature environments, suboptimal lighting conditions, diminutive scale targets themselves. To address this persistent challenge, MarineYOLO network is introduced. This approach involves refining C2f module into EC2f module, alongside integration Efficient Multi-scale Attention (EMA) YOLOv8. Additionally, Convolutional Block Module (CBAM) introduced to further refine Feature Pyramid Network (FPN), facilitating enhanced feature extraction pertinent Furthermore, CIoU replaced with Wise-IoU augment precision stability target localization. Experimental findings demonstrate that achieves an average (AP) 78.5% on RUOD dataset 88.1% URPC dataset, marking improvements 12.2% 16.8%, respectively, compared YOLOv8n. As emerging paradigm harbors significant potential both practical applications scholarly endeavors, furnishing efficacious remedy associated settings.

Language: Английский

A systematic review and analysis of deep learning-based underwater object detection DOI

Shubo Xu,

Minghua Zhang, Wei Song

et al.

Neurocomputing, Journal Year: 2023, Volume and Issue: 527, P. 204 - 232

Published: Jan. 11, 2023

Language: Английский

Citations

127
Experimental verification of ultra-broadband vibration reduction of underwater vehicle pressure-resisting shells using acoustic black holes DOI
Nansha Gao, Zhicheng Zhang, Yiting Li

et al.

Thin-Walled Structures, Journal Year: 2025, Volume and Issue: unknown, P. 113118 - 113118

Published: Feb. 1, 2025

Language: Английский

Citations

2
Dynamic YOLO for small underwater object detection DOI Creative Commons
Jie Chen, Meng Joo Er

Artificial Intelligence Review, Journal Year: 2024, Volume and Issue: 57(7)

Published: June 6, 2024

Abstract The practical application of object detection inevitably encounters challenges posed by small objects. In underwater detection, a crucial method for marine exploration, the presence objects in environments significantly hampers performance detection. this paper, dynamic YOLO detector is proposed as solution to alleviate problem. Specifically, light-weight backbone network first constructed based on deformable convolution v3, with some specialized designs Secondly, unified feature fusion framework channel-wise, scale-wise, and spatial-aware attention fuse maps from different scales. This particularly critical detecting since it allows us fully exploit enhanced capabilities offered our network. Finally, simple but effective head designed handle conflict between classification localization disentangling aligning two tasks. Extensive experiments are conducted benchmark datasets demonstrate effectiveness model. Without bells whistles, outperforms recent state-of-the-art methods large margin $$+\,0.8$$ + 0.8 AP $$+\,1.8$$ 1.8 $$\text {AP}_{S}$$ AP S DUO dataset. Experimental results Pascal VOC MS COCO also superiority method. At last, ablation studies dataset validate efficiency each design YOLO. Source code will be available at https://github.com/chenjie04/Dynamic-YOLO .

Language: Английский

Citations

14
Underwater Target Detection Using Deep Learning: Methodologies, Challenges, Applications, and Future Evolution DOI Creative Commons
Anwar Khan, Mostafa M. Fouda, Dinh‐Thuan Do

et al.

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 12618 - 12635

Published: Jan. 1, 2024

This paper provides a study of the latest target (object) detection algorithms for underwater wireless sensor networks (UWSNs). To ensure selection and state-of-the-art algorithms, only developed in last seven years are taken into account that not entirely addressed by existing surveys. These classified based on their architecture methodologies operation applications described helpful diverse set applications. The merits demerits also to improve performance future investigation. Moreover, comparative analysis is given further an insight various enhancement. A depiction publication count over decade (2023-2013) using IEEE database knowing application trend. Finally, challenges associated with highlighted research paradigms identified. conducted providing thorough feasibility defined strategies

Language: Английский

Citations

12
An advanced AI-based lightweight two-stage underwater structural damage detection model DOI

Xijun Ye,

Kanhui Luo,

Hanmin Wang

et al.

Advanced Engineering Informatics, Journal Year: 2024, Volume and Issue: 62, P. 102553 - 102553

Published: May 2, 2024

Language: Английский

Citations

12
HTDet: A Hybrid Transformer-Based Approach for Underwater Small Object Detection DOI Creative Commons
Gangqi Chen, Zhaoyong Mao, Kai Wang

et al.

Remote Sensing, Journal Year: 2023, Volume and Issue: 15(4), P. 1076 - 1076

Published: Feb. 16, 2023

As marine observation technology develops rapidly, underwater optical image object detection is beginning to occupy an important role in many tasks, such as naval coastal defense aquaculture, etc. However, the complex environment, images captured by imaging system are usually severely degraded. Therefore, how detect objects accurately and quickly under conditions a critical problem that needs be solved. In this manuscript, novel framework for based on hybrid transformer network proposed. First, lightweight transformer-based presented can extract global contextual information. Second, fine-grained feature pyramid used overcome issues of feeble signal disappearance. Third, test-time-augmentation method applied inference without introducing additional parameters. Extensive experiments have shown approach we proposed able small efficient effective way. Furthermore, our model significantly outperforms latest advanced detectors with respect both number parameters mAP considerable margin. Specifically, detector baseline 6.3 points, reduced 28.5 M.

Language: Английский

Citations

20
Window-based transformer generative adversarial network for autonomous underwater image enhancement DOI

Mehnaz Ummar,

Fayaz Ali Dharejo, Basit Alawode

et al.

Engineering Applications of Artificial Intelligence, Journal Year: 2023, Volume and Issue: 126, P. 107069 - 107069

Published: Sept. 9, 2023

Language: Английский

Citations

18
Optimization and Application of Improved YOLOv9s-UI for Underwater Object Detection DOI Creative Commons
Wei Pan,

Jiabao Chen,

Bangjun Lv

et al.

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(16), P. 7162 - 7162

Published: Aug. 15, 2024

The You Only Look Once (YOLO) series of object detection models is widely recognized for its efficiency and real-time performance, particularly under the challenging conditions underwater environments, characterized by insufficient lighting visual disturbances. By modifying YOLOv9s model, this study aims to improve accuracy capabilities detection, resulting in introduction YOLOv9s-UI model. proposed model incorporates Dual Dynamic Token Mixer (D-Mixer) module from TransXNet feature extraction capabilities. Additionally, it integrates a fusion network design LocalMamba network, employing channel spatial attention mechanisms. These modules effectively guide process, significantly enhancing while maintaining model’s compact size only 9.3 M. Experimental evaluation on UCPR2019 dataset shows that has higher recall than existing as well excellent performance. This improves ability target introducing advanced meets portability requirements provides more efficient solution detection.

Language: Английский

Citations

6
Lightweight underwater object detection based on image enhancement and multi-attention DOI
Tian Tian, Jixiang Cheng, Dan Wu

et al.

Multimedia Tools and Applications, Journal Year: 2024, Volume and Issue: 83(23), P. 63075 - 63093

Published: Jan. 10, 2024

Language: Английский

Citations

5
A Survey of Seafloor Characterization and Mapping Techniques DOI Creative Commons
Gabriel Loureiro, André Dias, José Almeida

et al.

Remote Sensing, Journal Year: 2024, Volume and Issue: 16(7), P. 1163 - 1163

Published: March 27, 2024

The deep seabed is composed of heterogeneous ecosystems, containing diverse habitats for marine life. Consequently, understanding the geological and ecological characteristics seabed’s features a key step many applications. majority approaches commonly use optical acoustic sensors to address these tasks; however, each sensor has limitations associated with underwater environment. This paper presents survey main techniques trends related characterization, highlighting in three tasks: classification, detection, segmentation. bibliography categorized into four approaches: statistics-based, classical machine learning, object-based image analysis. differences between are presented, challenges sea research potential directions study outlined.

Language: Английский

Citations

5