MAS-YOLOv11: An Improved Underwater Object Detection Algorithm Based on YOLOv11

To address the challenges of underwater target detection, including complex background interference, light attenuation, severe occlusion, and overlap between targets, as well as the wide-scale variation in objects, we propose MAS-YOLOv11, an improved model integrating three key enhancements: First,...

Full description

Saved in:
Bibliographic Details
Main Authors: Yang Luo, Aiping Wu, Qingqing Fu
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Sensors
Subjects:
Online Access:https://www.mdpi.com/1424-8220/25/11/3433
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850158817992507392
author Yang Luo
Aiping Wu
Qingqing Fu
author_facet Yang Luo
Aiping Wu
Qingqing Fu
author_sort Yang Luo
collection DOAJ
description To address the challenges of underwater target detection, including complex background interference, light attenuation, severe occlusion, and overlap between targets, as well as the wide-scale variation in objects, we propose MAS-YOLOv11, an improved model integrating three key enhancements: First, we introduce the C2PSA_MSDA module, which integrates multi-scale dilated attention (MSDA) into the C2PSA module of the backbone, enhancing multi-scale feature representation via dilated convolutions and cross-scale attention. Second, an adaptive spatial feature fusion detection head (ASFFHead) replaces the original head. By employing learnable spatial weighting parameters, ASFFHead adaptively fuses features across different scales, significantly improving the robustness of multi-scale object detection. Third, we introduce a Slide Loss function with dynamic sample weighting to enhance hard sample learning. By mapping the loss weights nonlinearly to detection confidence, this mechanism effectively enhances the overall detection accuracy. The experimental results demonstrate that the improved model yields significant performance advancements on the DUO dataset: the recall rate is enhanced by 3.7%, the F1-score is elevated by 3%, and the mAP@50 and mAP@50-95 attain values of 77.4% and 55.1%, respectively, representing increases of 3.5% and 3.3% compared to the baseline model. Furthermore, the model achieves an mAP@50 of 76% on the RUOD dataset, which further corroborates its cross-domain generalization capability.
format Article
id doaj-art-fe85c2be624848f0bbb4fe16eabc2696
institution OA Journals
issn 1424-8220
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Sensors
spelling doaj-art-fe85c2be624848f0bbb4fe16eabc26962025-08-20T02:23:45ZengMDPI AGSensors1424-82202025-05-012511343310.3390/s25113433MAS-YOLOv11: An Improved Underwater Object Detection Algorithm Based on YOLOv11Yang Luo0Aiping Wu1Qingqing Fu2School of Electronic Information and Electrical Engineering, Yangtze University, Jingzhou 434023, ChinaSchool of Computing Science and Artificial Intelligent, Suzhou City University, Suzhou 215104, ChinaSchool of Electronic Information and Electrical Engineering, Yangtze University, Jingzhou 434023, ChinaTo address the challenges of underwater target detection, including complex background interference, light attenuation, severe occlusion, and overlap between targets, as well as the wide-scale variation in objects, we propose MAS-YOLOv11, an improved model integrating three key enhancements: First, we introduce the C2PSA_MSDA module, which integrates multi-scale dilated attention (MSDA) into the C2PSA module of the backbone, enhancing multi-scale feature representation via dilated convolutions and cross-scale attention. Second, an adaptive spatial feature fusion detection head (ASFFHead) replaces the original head. By employing learnable spatial weighting parameters, ASFFHead adaptively fuses features across different scales, significantly improving the robustness of multi-scale object detection. Third, we introduce a Slide Loss function with dynamic sample weighting to enhance hard sample learning. By mapping the loss weights nonlinearly to detection confidence, this mechanism effectively enhances the overall detection accuracy. The experimental results demonstrate that the improved model yields significant performance advancements on the DUO dataset: the recall rate is enhanced by 3.7%, the F1-score is elevated by 3%, and the mAP@50 and mAP@50-95 attain values of 77.4% and 55.1%, respectively, representing increases of 3.5% and 3.3% compared to the baseline model. Furthermore, the model achieves an mAP@50 of 76% on the RUOD dataset, which further corroborates its cross-domain generalization capability.https://www.mdpi.com/1424-8220/25/11/3433underwater object detectionYOLOv11attention mechanismdetection headloss functionmulti-scale feature learning
spellingShingle Yang Luo
Aiping Wu
Qingqing Fu
MAS-YOLOv11: An Improved Underwater Object Detection Algorithm Based on YOLOv11
Sensors
underwater object detection
YOLOv11
attention mechanism
detection head
loss function
multi-scale feature learning
title MAS-YOLOv11: An Improved Underwater Object Detection Algorithm Based on YOLOv11
title_full MAS-YOLOv11: An Improved Underwater Object Detection Algorithm Based on YOLOv11
title_fullStr MAS-YOLOv11: An Improved Underwater Object Detection Algorithm Based on YOLOv11
title_full_unstemmed MAS-YOLOv11: An Improved Underwater Object Detection Algorithm Based on YOLOv11
title_short MAS-YOLOv11: An Improved Underwater Object Detection Algorithm Based on YOLOv11
title_sort mas yolov11 an improved underwater object detection algorithm based on yolov11
topic underwater object detection
YOLOv11
attention mechanism
detection head
loss function
multi-scale feature learning
url https://www.mdpi.com/1424-8220/25/11/3433
work_keys_str_mv AT yangluo masyolov11animprovedunderwaterobjectdetectionalgorithmbasedonyolov11
AT aipingwu masyolov11animprovedunderwaterobjectdetectionalgorithmbasedonyolov11
AT qingqingfu masyolov11animprovedunderwaterobjectdetectionalgorithmbasedonyolov11