Water Surface Garbage Detection and Autonomous Navigation Cleaning Technology Based on an Enhanced YOLO Visual Algorithm
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
This study addresses the engineering challenges associated with the removal of small floating debris from water surfaces, with a core focus on algorithmic customization for specific scenarios, deployment on low-computational-power hardware, and system-level integration and optimization. To address the challenges posed by high surface reflectivity and the prevalence of small targets in aquatic environments, the spatial-channel attention mechanism is specifically adapted and optimized for this application context. The enhanced model is integrated within the YOLO framework and successfully deployed on the STM32H743 embedded platform. Furthermore, the system incorporates multi-sensor fusion for navigation and an expandable mechanical structure, enabling a fully integrated closed-loop workflow encompassing detection, navigation, and debris collection. To enhance the detection performance for small-sized objects and garbage in complex environmental conditions, the study implements targeted optimizations to the object detection algorithm by integrating an attention mechanism and training on a custom-built dataset that closely reflects real-world scenarios. The final algorithm maintains a lightweight architecture and high-speed inference capability while ensuring a high average recognition accuracy.
Fields of Study & Subjects
Advanced Neural Network Applications
Water Quality Monitoring Technologies
Industrial Vision Systems and Defect Detection
Publication details
- Journal / source
- International Conference on Electrical Automation and Artificial Intelligence (ICEAAI)
- Pages
- 745–751
- Publication date
- 16 January 2026
- Language
- EN