EARLY DETECTION OF FISH DISEASES BY ANALYSING QUALITY OF WATER USING MACHINE LEARNING ALGORITHM

Abstract

Diseases that affect fish in aquaculture pose a substantial threat to the industry's ability to provide enough nutrition. The lack of sufficient infrastructure makes it difficult to identify diseased fish in aquaculture at an early stage, making it difficult to figure out whether or not fish have been infected. A necessary step in the fight against the further spread of illness is the prompt and accurate identification of diseased fish. Because salmon aquaculture is the world's fastest-growing system for the production of food, and because salmon aquaculture accounts for 70 percent of the market (2.5 million tonnes), the goal of this investigation is to discover the illness that affects salmon fish in aquaculture. We are able to detect the diseased fishes caused by a variety of pathogens thanks
to a synergistic partnership between faultless image processing and machine learning method. This work is separated into two distinct parts. In the preliminary stage, picture pre-processing and segmentation have been used to, respectively, lessen the amount of noise
and amplify the appearance of the image. In the second part of our analysis, we use an approach for machine learning called Support Vector Machine (SVM) with a kernel function to categorise illnesses and then extract the characteristics that are involved in each disease. The processed pictures from the first section were run through this support vector machine (SVM) model. After that, we harmonise a full experiment using the suggested combination of approaches on the salmon fish picture dataset that was used to investigate the fish illness. This work has been presented on a new dataset that includes both cases of image augmentation and
cases where it was not used. The findings have led us to the conclusion that the SVM we have been using operates well, achieving an accuracy of 91.42 and 94.12 percent, respectively, with and without augmentation.