Objective

The identification of marine objects, particularly underwater species, has traditionally relied on manual methods, visual observation, and limited automated techniques. Current approaches often involve time-consuming and labor-intensive processes, such as human divers visually cataloging species or utilizing basic image recognition algorithms. However, these methods are prone to inaccuracies and lack the precision required for comprehensive marine species monitoring. The inherent challenges of underwater environments, such as varying lighting conditions and complex backgrounds, further exacerbate the limitations of existing methodologies.

In response to these challenges, the application of computer vision, particularly advanced object detection models like YOLOv8, emerges as a transformative solution. By leveraging deep learning and sophisticated algorithms, computer vision offers the potential to enhance the accuracy and efficiency of marine object identification. This study focuses on harnessing the power of YOLOv8 to address the shortcomings of current methods, providing a more accurate and precise approach to underwater marine species identification.

The model is trained on a meticulously curated dataset featuring five distinct marine species: fish, eel, jellyfish, lobster, and lionfish. Each of these classes represents a critical component of marine ecosystems, and the accurate identification of these species is essential for marine biologists, ecologists, and conservationists.