An Approach for Fish Disease Recognition

Abstract

Fisheries is a prominent sector in Bangladesh. It has giant contribution in our economy. There are many people involved in this sector for their livelihood in this country. There are lots of principle invested in fisheries. This sector has also some barriers. Infected diseases are the prime culprit for fish cultivation. These diseases are created havoc situation for fish cultivation. Infected diseases apparent many problems in fish cultivation, such as decrease of production, deficit of protein, wracked of cultivator’s investment. So, cultivators need proper steps from heinous hands of these diseases. Traditional evaluation of these diseases are very difficult task for cultivators according to time and proper management. Through computer vision approach technological identification of these diseases is becoming more effective way day by day. So in our work we expose a sophisticated way to identify the diseases by which fish are infected. Epizootic Ulcerative Syndrome (EUS) and Tail and Fin Rot are most common infected diseases of fish cultivation in our country. This proposed method would be helped to the fisheries sector in Bangladesh to cultivate fish for effectively and probably increasing fish production by measuring proper way after technological identification of diseases. K-means clustering is being used for image segmentation after image preprocessing system is implemented. Ten particular traits are omitted in our thesis. Multiple classification techniques are used for classification. Random Forest algorithm gains almost 80.62% proficiency.