DETECTION OF GIANT CLAM USING YOLOv7 FOR PRECISION BIODIVERSITY

Abstract

Giant clams, vital to coral reef ecosystems, have experienced significant population declines due to overfishing and habitat destruction. Monitoring these populations poses challenges, as traditional scuba diving surveys are labor-intensive and limited in scope. In this study, a YOLOv7 object detection algorithm was developed and implemented on an underwater remotely operated vehicle (ROV) to address these challenges. A comprehensive dataset of giant clam images was created, enabling the training and evaluation of the model. The system achieved an average accuracy of 79.65% in detecting and counting giant clams across six different underwater scenes, demonstrating its potential for automated population assessment. Accuracy could be improved by expanding the dataset with higher-quality images, enhancing reliability for real-world applications. This approach provides a promising tool for advancing marine life recognition and supporting conservation strategies, particularly in monitoring and studying giant clam populations in marine protected areas.