BIOINFORMATICS FISH MORPHOLOGY DETECTION SYSTEM DEEP LEARNING METHOD USING MICROCONTROLLER-BASED CONVOLUTIONAL NEURAL NETWORK ALGORITHM

Abstract

Fish are members of poikilothermic vertebrates that live in water and breathe with gills, fish are the most diverse group of vertebrates with more than 27,000 species worldwide, until now fish are generally consumed fresh. Fish can be processed into various products such as dried fish, fish jerky, shredded fish, fish crackers, salted fish, fish balls and fish blood flour as plant fertilizer and fish feed. Indonesia is the world's largest exporter of fisheries at 937,201.14 tons and national fish consumption is 54.9 kg/capita (2021). Artificial Intelligence (AI) is a technique used to mimic the intelligence possessed by living and non-living things to solve a problem. Deep Learning is a technique used in the development of neural networks that uses certain techniques to accelerate the learning process. Previous research used image processing-based to detect fish diseases. Bioinformatics is the study/application of computational techniques to manage and analyze biological information, including the application of mathematical, statistical, and informatics methods to solve biological problems and facilitate the identification and classification of fish species as part of biological methods for learning. The purpose of this research is to create a bioinformatics system for fish species detection using deep learning method with multi sensors (eyes, smell, color, texture, gills) based on microcontroller control system (arduino) and smartphone. The purpose of this research is to analyze and make a prototype tool to detect the type of skin color and fish eyes based on fish biology (bioinformatics) with deep learning method using microcontroller-based Convolutional Neural Network algorithm. The sample test verification result is a sampling error of 25% for FCL and 20% for FCE, so it can still be used as a tool for identifying fish species.