Real-Time Clustering of Seagrass Age Categories Using Deep Learning and Unsupervised Machine Learning

Abstract

Seagrass ecosystems play a vital role in maintaining marine biodiversity and ecological balance, making their monitoring and management essential. This study proposes a novel approach for real-time clustering of seagrass images into three distinct age categories—young, medium, and old—using deep learning and unsupervised machine learning techniques. We employ the VGG-16 convolutional neural network (CNN) for feature extraction from a dataset of 800 seagrass images, followed by K-means clustering to categorize them. Our methodology includes image preprocessing, VGG-16 model optimization for real-time processing, and feature extraction followed by K-means clustering. We evaluate the clustering results using metrics like silhouette score and Davies-Bouldin index, along with performance visualizations through ROC curves and confusion matrices. The findings demonstrate the effectiveness of our approach in capturing age-related patterns, providing a valuable tool for marine ecosystem management. The model achieved a silhouette score of 71% and a Davies-Bouldin index of 42%, indicating strong intra-cluster similarity and well-separated clusters. These results outperform traditional image-based classification methods, validating the robustness of our real-time clustering approach.