RTDETR-MARD: A Multi-Scale Adaptive Real-Time Framework for Floating Waste Detection in Aquatic Environments

Accurate and efficient detection of floating waste is crucial for environmental protection and aquatic ecosystem preservation, yet remains challenging due to environmental interference and the prevalence of small targets. To address these limitations, we propose a Multi-scale Adaptive Real-time Dete...

Full description

Saved in:
Bibliographic Details
Main Authors: Baoshan Sun, Haolin Tang, Liqing Gao, Kaiyu Bi, Jiabao Wen
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Journal of Marine Science and Engineering
Subjects:
Online Access:https://www.mdpi.com/2077-1312/13/5/996
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849327278759084032
author Baoshan Sun
Haolin Tang
Liqing Gao
Kaiyu Bi
Jiabao Wen
author_facet Baoshan Sun
Haolin Tang
Liqing Gao
Kaiyu Bi
Jiabao Wen
author_sort Baoshan Sun
collection DOAJ
description Accurate and efficient detection of floating waste is crucial for environmental protection and aquatic ecosystem preservation, yet remains challenging due to environmental interference and the prevalence of small targets. To address these limitations, we propose a Multi-scale Adaptive Real-time Detector (RTDETR-MARD) based on RT-DETR that introduces three key innovations for improved floating waste detection using unmanned surface vessels (USVs). First, our hierarchical multi-scale feature integration leverages the gather-and-distribute mechanism to enhance feature aggregation and cross-layer interaction. Second, we develop an advanced feature fusion module incorporating feature alignment, Information Fusion, information injection, and Scale Sequence Feature Fusion components to ensure precise spatial alignment and semantic consistency. Third, we implement the Wise-IoU loss function to optimize localization accuracy through high-quality anchor supervision. Extensive experiments demonstrate the framework’s effectiveness, achieving state-of-the-art performance of 86.6% mAP50 at 96.8 FPS on the FloW dataset and 49.2% mAP50 at 107.5 FPS on our custom water surface waste dataset. These results confirm RTDETR-MARD’s superior accuracy, real-time capability, and robustness across diverse environmental conditions, making it particularly suitable for practical deployment in ecological monitoring systems where both speed and precision are critical requirements.
format Article
id doaj-art-e2b269bf31e2432daac929f079d61d1d
institution Kabale University
issn 2077-1312
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Journal of Marine Science and Engineering
spelling doaj-art-e2b269bf31e2432daac929f079d61d1d2025-08-20T03:47:54ZengMDPI AGJournal of Marine Science and Engineering2077-13122025-05-0113599610.3390/jmse13050996RTDETR-MARD: A Multi-Scale Adaptive Real-Time Framework for Floating Waste Detection in Aquatic EnvironmentsBaoshan Sun0Haolin Tang1Liqing Gao2Kaiyu Bi3Jiabao Wen4School of Computer Science and Technology, Tiangong University, Tianjin 300387, ChinaSchool of Computer Science and Technology, Tiangong University, Tianjin 300387, ChinaSchool of Computer Science and Technology, Tiangong University, Tianjin 300387, ChinaSchool of Computer Science and Technology, Tiangong University, Tianjin 300387, ChinaSchool of Electrical and Information Engineering, Tianjin University, Tianjin 300052, ChinaAccurate and efficient detection of floating waste is crucial for environmental protection and aquatic ecosystem preservation, yet remains challenging due to environmental interference and the prevalence of small targets. To address these limitations, we propose a Multi-scale Adaptive Real-time Detector (RTDETR-MARD) based on RT-DETR that introduces three key innovations for improved floating waste detection using unmanned surface vessels (USVs). First, our hierarchical multi-scale feature integration leverages the gather-and-distribute mechanism to enhance feature aggregation and cross-layer interaction. Second, we develop an advanced feature fusion module incorporating feature alignment, Information Fusion, information injection, and Scale Sequence Feature Fusion components to ensure precise spatial alignment and semantic consistency. Third, we implement the Wise-IoU loss function to optimize localization accuracy through high-quality anchor supervision. Extensive experiments demonstrate the framework’s effectiveness, achieving state-of-the-art performance of 86.6% mAP50 at 96.8 FPS on the FloW dataset and 49.2% mAP50 at 107.5 FPS on our custom water surface waste dataset. These results confirm RTDETR-MARD’s superior accuracy, real-time capability, and robustness across diverse environmental conditions, making it particularly suitable for practical deployment in ecological monitoring systems where both speed and precision are critical requirements.https://www.mdpi.com/2077-1312/13/5/996real-time object detectionfloating waste monitoringadaptive feature aggregationmulti-scale feature fusion
spellingShingle Baoshan Sun
Haolin Tang
Liqing Gao
Kaiyu Bi
Jiabao Wen
RTDETR-MARD: A Multi-Scale Adaptive Real-Time Framework for Floating Waste Detection in Aquatic Environments
Journal of Marine Science and Engineering
real-time object detection
floating waste monitoring
adaptive feature aggregation
multi-scale feature fusion
title RTDETR-MARD: A Multi-Scale Adaptive Real-Time Framework for Floating Waste Detection in Aquatic Environments
title_full RTDETR-MARD: A Multi-Scale Adaptive Real-Time Framework for Floating Waste Detection in Aquatic Environments
title_fullStr RTDETR-MARD: A Multi-Scale Adaptive Real-Time Framework for Floating Waste Detection in Aquatic Environments
title_full_unstemmed RTDETR-MARD: A Multi-Scale Adaptive Real-Time Framework for Floating Waste Detection in Aquatic Environments
title_short RTDETR-MARD: A Multi-Scale Adaptive Real-Time Framework for Floating Waste Detection in Aquatic Environments
title_sort rtdetr mard a multi scale adaptive real time framework for floating waste detection in aquatic environments
topic real-time object detection
floating waste monitoring
adaptive feature aggregation
multi-scale feature fusion
url https://www.mdpi.com/2077-1312/13/5/996
work_keys_str_mv AT baoshansun rtdetrmardamultiscaleadaptiverealtimeframeworkforfloatingwastedetectioninaquaticenvironments
AT haolintang rtdetrmardamultiscaleadaptiverealtimeframeworkforfloatingwastedetectioninaquaticenvironments
AT liqinggao rtdetrmardamultiscaleadaptiverealtimeframeworkforfloatingwastedetectioninaquaticenvironments
AT kaiyubi rtdetrmardamultiscaleadaptiverealtimeframeworkforfloatingwastedetectioninaquaticenvironments
AT jiabaowen rtdetrmardamultiscaleadaptiverealtimeframeworkforfloatingwastedetectioninaquaticenvironments