Abstract:
Despite the fact that fish is a major source of food production in the world, fish infections continue to threaten the productivity of farms, economic stability, and food security, especially among countries such as Bangladesh that rely on fish farming as a major source of livelihood. Traditional methods of detecting disease make use of laboratory tests and physical inspection, which is tedious, inconsistent, and often inaccessible to small farmers. In order to address these limitations, this paper presents an automated deep learning-based system referred to as "YOLO-Based Fish Disease Detection: A Smart Lifeline to Aquaculture Farmers. A well-chosen sample of 1,406 raw photos which were further augmented to 7,710 photos was collected and grouped into seven categories, including healthy and sick fish. This dataset was followed by training and evaluating a YOLO trained to permit the precise localization of bounding boxes and also identification of multiple classes in real-time. Experimental results prove that the model is suitable to be employed in practical farms, to achieve high detection accuracy and stable work in contrast to a different light and climatic conditions. It is an automated process of assisting aquaculture farmers in effective decision making, enhances prompt intervention of diseases and reduces the diagnostic delay significantly. The work propels technology-hackable aquaculture approaches and seals existing gaps in the current real-time monitoring of fish diseases through a simple AI-based solution and fast and accurate.
