Underwater Garbage Detection Using YOLOv8 Model

Abstract: Effective machine-learning algorithms are vital for the successful navigation of underwater environments and the intelligent recognition of objects in murky waters. The advancement of modern society has led to increased pollution in marine ecosystems, particularly in oceans, rivers, and lakes, which threatens our precious water resources. Despite existing environmental regulations, solid waste, including refuse and debris, continues to be directly dumped into the ocean, negatively affecting the survival and health of marine life. Consequently, it is imperative to employ suitable methods for the precise detection and analysis of features in these specific environments. In this study, we have utilized the YoloV8 algorithm for underwater waste detection, utilizing a dataset comprising 5096 images from various categories. These categories encompass items such as masks, metal cans, glass bottles, gloves, plastic bags, and tires, captured across a range of distinct underwater settings. This research also tackles the escalating problem of underwater waste in oceans and seas by identifying debris in underwater imagery.

Keywords: Object Detection, Deep Learning, Waste Detection, YOLOv8, Underwater Image, Marine Plastic Waste Detection.