Automated Freshwater Fish Species Classification using Deep CNN

Freshwater fish is considered a poor man’s protein supplement as they are easily available in lakes, rivers, natural ponds, paddy fields, beels, and fisheries. There various freshwater species that resemble each other, making it difficult to classify them by their external appearance. Manual identification always needs expertise so, erroneous. Recently, computer vision along with deep learning plays significant role underwater classification research where the number of under investigation limited maximum eight (8). In this article, we choose deep-learning architectures, AlexNet Resnet-50, 20 indigenous fresh-water from North-Eastern parts India. The two models fine-tuned for training validation collected data. performance these networks evaluated based on overall accuracy, precision, recall rate. This paper reports best rate 100% at 0.001 Resnet-50 model our own dataset benchmark Fish-Pak dataset. Comprehensive empirical analysis has proved an increasing Weight Bias rate, loss incurred classifier also increases.