An improved YOLOv8-based method for litter identification in drone shoreline images

Key Points

• Accuracy improves by 2.1% using the YOLOv8-b model, focusing on identifying shoreline litter effectively.
• The model tests reveal a recall rate increase of 1.9%, assuring better detection of four litter categories.
• Applications of the improved YOLOv8-b model in Tangshan Bay show over 90% precision in identifying various litter types.
• Further development supports business monitoring requirements for shoreline litter management.

Abstract