A Marine Debris Detection Framework for Ocean Robots via Self-Attention Enhancement and Feature Interaction Optimization

2026-05-08

AI Summary

This work addresses the challenge of accurately detecting marine debris in low-quality underwater images characterized by hazy conditions, complex backgrounds, and small target sizes. To this end, the authors propose YOLO-MD, a novel detection framework that integrates a dual-branch convolution-enhanced self-attention module (DB-CASA), a lightweight shift operation, and a specially designed SFG-Loss function, complemented by a dynamic sample reweighting strategy. These components collectively enhance multi-scale fine-grained feature extraction and facilitate effective spatial-channel feature interaction. Evaluated on the UODM dataset, YOLO-MD achieves a precision of 0.875, an F1-score of 0.822, and an mAP50 of 0.849, outperforming current state-of-the-art methods. Furthermore, the model has been successfully deployed on real-world marine robotic edge devices, demonstrating significantly improved robustness and accuracy in degraded visual environments.

Abstract

Marine debris detection for ocean robot is crucial for ecological protection, yet performance is often degraded by low-quality images with blur, complex backgrounds, and small targets. To address these challenges, we propose YOLO-MD, an enhanced YOLO-based detection framework. A Dual-Branch Convolutional Enhanced Self-Attention (DB-CASA) module is designed to strengthen spatial-channel interactions, improving feature representation in degraded images. Additionally, a lightweight shift-based operation is introduced to enhance fine-grained feature extraction for objects of varying scales while maintaining parameter efficiency. We further propose SFG-Loss to mitigate class imbalance and optimization instability via dynamic sample reweighting. Experiments on the UODM dataset demonstrate that YOLO-MD achieves 0.875 precision, 0.822 F1-score, and 0.849 mAP50, outperforming the latest state-of-the-art methods. The effectiveness of this method has also been verified through real-world robotic edge deployment experiments.

Problem

• marine debris detection
• low-quality images
• complex backgrounds
• small targets
• ocean robots

Innovation

• Self-Attention Enhancement
• Feature Interaction Optimization
• Lightweight Shift Operation
• Dynamic Sample Reweighting
• Marine Debris Detection