Abstract: Plastic Identification Underwater addresses the pressing issue of plastic pollution in our oceans by developing a novel underwater plastic object detection system using the YOLOv8 architecture. The model, trained on a diverse dataset from Roboflow, achieved an accuracy rate of 87% in detecting submerged plastic objects. A Flask web application was created to enable users to perform object detection on live video streams and specific paths, while integration with a mobile application allows real-time detection using mobile cameras underwater. This comprehensive solution aims to enhance marine conservation efforts by facilitating the identification and removal of underwater plastic debris.

Keywords: Underwater plastic detection, YOLOv8, Marine conservation, Real-time detection

| DOI: 10.17148/IARJSET.2024.11821