Plastic waste over the ocean: an approach to surface recognition.

Shah, Faisal Mehmood; Fough, Faranak; Shah, Zohaib Mehmood; Fough, Nazila; Zhao, Yafan

Abstract

To address the challenges of low detection accuracy and slow recognition of floating objects on water surfaces, an enhanced YOLOv5 algorithm has been developed. This improved algorithm incorporates a coordinate attention mechanism to enhance spatial recognition and employs the GIOU loss function with updated anchor matching for quicker decision-making. The new method achieved a 97.39% mAP, marking a 15.28% improvement over the original YOLOv5, demonstrating its effectiveness.

Citation

SHAH, F.M., FOUGH, F., SHAH, Z.M., FOUGH, N. and ZHAO, Y. 2024. Plastic waste over the ocean: an approach to surface recognition. In Proceedings of the 31st IEEE (Institute of Electrical and Electronics Engineers) International conference on electronics, circuits, and systems (ICECS 2024), 18-20 November 2024, Nancy, France. Piscataway: IEEE [online], article 10849157. Available from: https://doi.org/10.1109/ICECS61496.2024.10849157

• Presentation Conference Type: Conference Paper (published)
• Conference Name: 31st IEEE (Institute of Electrical and Electronics Engineers) International conference on electronics, circuits, and systems (ICECS 2024)
• Start Date: Nov 18, 2024
• End Date: Nov 20, 2024
• Acceptance Date: Jul 29, 2024
• Online Publication Date: Jan 28, 2025
• Publication Date: Dec 31, 2025
• Deposit Date: Mar 11, 2025
• Publicly Available Date: Mar 11, 2025
• Print ISSN: 2994-5755
• Electronic ISSN: 2995-0589
• Publisher: Institute of Electrical and Electronics Engineers (IEEE)
• Peer Reviewed: Peer Reviewed
• Article Number: 10849157
• DOI: https://doi.org/10.1109/icecs61496.2024.10849157
• Keywords: Pattern recognition; YOLOv5; Sparse features; Attention mechanism; CFN
• Public URL: https://rgu-repository.worktribe.com/output/2709573