Search

In natural water such as the sea, and rivers, people produce trash and debris that are very common. Removing this trash and debris is necessary for the well-being of marine and human life. It is difficult to detect and map debris under the water because of its peculiar characteristics. In order to remove submerged debris from the sea we should have a detailed knowledge about the debris. Hence, an autonomous system is required for classification and detection of submerged marine debris. This paper proposes to classify marine debris by its material and object, thereby allowing us to determine what kind of damage the debris poses to the environment. For multiclass identification, the suggested research employs neural network architecture built on the You Only Look Once (YOLO) framework. The performance evaluation showed that the YOLOv5 model gave a mAP of 72% while classifying and detecting 5 classes of debris. This model, when used appropriately in an autonomous underwater vehicle can prove to be effective in creating a sustainable underwater environment. This study also examines the influence of marine debris on the ocean environment using various factors such as degradation time, ingestion by marine life, and so on. [ABSTRACT FROM AUTHOR]