Classification of Fish Species Using CNN

In this work, classification of fish species (10 types) is performed using Convolutional Neural Network (CNN). Despite its industrial and agricultural utility, fish identification is currently an extremely complicated and challenging process. Certain difficulties in the accuracy and classification of fish involve distortion, noise, segmentation error, blur, and compression. A variety of techniques have been commonly used, including K Nearest Neighbor (KNN), K Mean Clustering and Support Vector Machine (SVM). Each methodology has inherent limits that restrict the accuracy of the classification task. This paper proposes a methodology based on a Convolutional Neural Network to eliminate drawbacks of some current methods and to improve the classification of fish species. The dataset for this work is collected from the fish4knowledge portal.

Keywords: Fish Species, Convolutional Neural Network, Classification, Support Vector Machine (SVM).