A Substantial Deep Learning Approach for Classification of Local and Coastal Fish

Abstract

Fish is a popular food all around the world Because of its excellent nutritional content. Furthermore, fish is low in fat and high in protein. The nutritional value of various fish varies. Fish are essential experimental animals in a variety of fields of biological and medical research. A solid foundation for understanding the more adaptable behavior of higher vertebrates has been established by research on fish. This article focused on the classification of two types of fish: local and coastal fish. This will aid in identifying fish, and this article will also provide knowledge of numerous fish species identifications, allowing researchers to study the nutritional value of fish. The local and coastal fish categories contain twelve different fish species: Catla, Cyprinus Carpio, Grass Carp, Mori, Rohu, Silver, Black Sea Sprat, Gilt Head Bream, Red Sea Bream, Horse Mackerel, Sea Bass, and Trout. Moreover, there are 13,176 fish shots in the dataset used in this article. In addition, to identify the species, fish are labeled with unique integer values. A deep learning based approach has been applied to classify the fish species in this article. A Convolutional Neural Network (CNN) technique has been used in this research work as CNN provides high-quality performance in the field of image segmentation. Hence, the proposed model achieves a satisfactory result of 98.33%.