Fish Recognition Using Deep Neural Network

Abstract

Fish recognition is the various essential factors of fishery studies applications, where a massive quantity of facts is gathered quickly. Due to bad picture quality, uncontrollable objects, and the environment, in addition to the problems in getting consultant samples, underwater picture popularity poses particular challenges. The primary purpose of this study is to create a supervised feature learning-based fish recognition framework. The required data is provided for further analysis based on medical and fish market usage. The system modules in this work are built using deep neural networks. Neural networks will increase accuracy in a variety of circumstances involving input photographs and targets. Experiments demonstrate that the suggested framework achieves great accuracy while balancing high uncertainty and sophistication on both sides: Public and self-collected underwater fish photos. Finally, the recognized fish type and medicinal uses are called out by utilizing voice instructions on MATLAB plateform.