Formalin Fish Detection System Based on Digital Image Processing

Fish is an Indonesian marine commodity that is mostly consumed by humans and exported. Relative the price is cheap and high nutritional content, but fish are easily damaged, especially in tropical conditions. Fish quality declines very quickly, so they are easy to spoil. They prevent spoilage is preservation. Therefore, preservation is often ignored by irresponsible people by preserving with dangerous chemicals that are still happening today. The circulation of the issue of formalin raises anxiety for the public as consumers. They choose fish based on their judgments and standards, consumers tend focus on the eyes and gills of fish, so the solution a mobile system application based on digital image processing with deep learning implementation uses the Convolutional Neural Network (CNN) algorithm, Mobilenet as the network architecture model. Based on the eye dataset, training accuracy have arrived at 100% and validation100%. The gill’s dataset training accuracy have arrived at 98% and validation 80%. Testing of the MobileNet architecture on the mobile application, eye’s dataset accuracy 100% and gill’s dataset accuracy 95%.