Underwater Fish Detection and Classification in Indian Waters Using E-dive

Abstract

The 70% of the world's world ’s population depends on seafood as its principal source of protein. This creates the unorganized and illegal fishing practices a menace to marine life. To support and conserve endangered species, the The study describes a method for automatically identifying and classifying various species of fish, includes dolphins, sharks, as well as others. Images captured with boat webcams are plagued by a variety of limitations, notably differing amounts of light intensity and opacity. The methodology that has been developed intends to help the researcher and environmentalist in analyzing images of fish acquired using the cameras on the boat, identifying them and classifying them into several types of fish according to their characteristics. The system adapts to shifts in lighting, brightness, etc. for the detection process. Three parts try to compensate the system's methodology. Throughout that phase, fish are detected in the image by scanning for areas that have a significant likelihood of harboring fish. The identification of the discovered fish into its species is the third phase. At this stage, the classifier model receives a segmented image of a fish and determines which species the newly discovered fish belongs to. In order to extract and evaluate the information, different Convolutional Neural Network designs are used throughout the recognition and classification phases. Each image is allocated a likelihood quotient between 0 to 1. It is the probability that the fish belongs to one of the nine key categories—black Sea Spart,Shrimp,Sea Bass, Red Mullet, Trout, Glit Head Bream, Striped Red Mullet, Horse Mackerel and Read Sea Bream