Object detection and classification in marine ecosystem using deep learning neural network / Muhammad Afiq Azman

The marine ecosystem is vital for maintaining ecological balance and biodiversity, serving as a habitat for countless species and supporting human livelihoods. This study explores the application of artificial intelligence (AI) and machine learning (ML) for the detection and classification of marine organisms using YOLOv8 and ResNet50 models. The primary objective is to develop and implement artificial intelligence (AI) and machine learning (ML) algorithms tailored to effectively identify within marine ecosystems. A comparative performance evaluation revealed that while YOLOv8 excels in object detection with high precision (0.85) and recall (0.83) due to its multiscale feature extraction capabilities, ResNet50 demonstrated higher overall accuracy (77%) in classification tasks. YOLOv8 outperforms in handling multiple objects in complex backgrounds, whereas ResNet50 struggles with multiple-class detection in single images, attributed to its architecture designed primarily for single-object classification. These findings highlight the complementary strengths of both models in advancing marine ecosystem analysis.