An Approach Towards Marine Animal Detection and Appreciation with Advanced Deep Learning Model Techniques

Abstract

Marine ecosystems are vital components of our planet, housing a diverse array of species. Monitoring and understanding these ecosystems are essential for conservation efforts and scientific research. This paper presents a novel approach to detecting and recognising marine animals using advanced deep learning models, specifically MobileNet and ResNet-50, in the context of underwater image analysis. In recent years, deep learning has made significant strides in computer vision, and its application to marine biology offers promising opportunities. MobileNet and ResNet-50 are chosen for their efficiency and accuracy, making them suitable for real-time deployment in underwater environments. The proposed system employs a two-step process: object detection and species recognition. Firstly, Mobile Net is utilized for object detection to locate marine animals in underwater images. Next, ResNet-50 is applied for fine-grained species recognition, classifying the detected animals into specific categories. The model is trained on a comprehensive dataset comprising diverse marine species to ensure robust performance. Our experiments demonstrate the effectiveness of the approach in accurately detecting and recognizing marine animals across various underwater conditions, including low visibility and different lighting conditions. The system's performance is evaluated based on detection accuracy, species classification accuracy, and computational efficiency. This research contributes to the field of marine biology by providing a reliable and efficient tool for monitoring and studying marine life. The proposed deep learning-based system can assist researchers, conservationists, and marine biologists in cataloguing and understanding marine ecosystems, ultimately supporting conservation efforts and advancing our knowledge of these critical environments.