Comparative Analysis of Jellyfish Classification: A Study Using YOLOv8 and Pre-trained Models

Jellyfish classification holds significant importance due to its implications in environmental monitoring, marine conservation, and public safety. Understanding the distribution and characteristics of jellyfish populations can provide valuable insights into the health of marine ecosystems, helping to manage fisheries, protect biodiversity, and mitigate potential ecological imbalances. Therefore, this study experimented with deep learning models for jellyfish classification, ranging from traditional convolutional neural networks (CNNs) such as DenseNet, InceptionV3, MobileNet, MobileNetV2, and NASNet Mobile and as well as the newly introduced YOLO architecture, YOLOv8. The dataset was comprised of six different types of jellyfish images. Notably, YOLOv8 stands out with an impressive accuracy of 99.5%, demonstrating its efficacy in accurately detecting and classifying jellyfish instances. Other models, such as DenseNet and MobileNet, also yield high accuracy scores, emphasizing the versatility of deep learning models in handling complex image classification tasks. The robust methodologies employed in training these models, combined with the achieved high accuracies, pave the way for enhanced understanding and monitoring of jellyfish populations, contributing to the sustainable management of marine ecosystems. In conclusion, jellyfish classification using YOLOv8 architecture is a crucial tool for marine researchers, environmentalists, and policymakers