An Overview on Machine Learning Techniques for Detection and Classification of Fish in Aquaculture

Abstract

Fish detection and classification are very important in marine biodiversity conservation, fisheries management, and aquaculture monitoring. This area has changed dramatically over the years with a general movement from manual observations and simple image processing techniques to complex AI models. This paper provides an analysis of the methodologies employed, demonstrating the shift from conventional techniques to feature-based machine learning with additional advancements such as deep learning techniques like Convolutional Neural Networks (CNN) and object detection frameworks. Specific problems related to underwater environments, like dataset limitations and recent progress in dealing with challenges like turbidity and low lighting, are discussed.

Topics & Concepts