International Research Journal of Education and Technology

The objective of this project is to develop an efficient and accurate system for detecting jellyfish in underwater environments using deep learning techniques. Massive jellyfish outbreaks pose serious risks to human safety and marine ecosystems, highlighting the urgent need for reliable detection methods. To address this, the project leverages optical imagery and a convolutional neural network (CNN)-based object detection approach.

Due to the scarcity of labeled jellyfish datasets, a novel dataset was created using a model- assisted labeling strategy, significantly minimizing the need for manual annotation. Based on this dataset, we propose an enhanced YOLOv11 model that incorporates the Global Attention Mechanism (GAM) and CoordConv modules to improve feature extraction and spatial awareness.

Experimental evaluations demonstrate that the proposed model outperforms several state-of-the- art detection frameworks in terms of accuracy and robustness. The results indicate the system's potential for real-time jellyfish detection, contributing to improved marine safety and ecosystem monitoring.