Klasifikasi Kesegaran Ikan Menggunakan Citra Mata dengan Convolutional Neural Network Arsitektur VGG-16

Abstract

Sea fish are the most widely consumed type of fish by households in Indonesia, serving as an important source of protein for the body. According to Matondang's (2022) study titled "Comparison of Protein Content in Freshwater Fish and Sea Fish," the protein content in sea fish is higher than in freshwater fish, making high-quality fish highly beneficial for the body. The fishing industry plays a crucial role in food supply, especially in maritime countries like Indonesia. The freshness of sea fish, as the main protein source for many households, significantly determines its quality and safety for consumption. Freshness affects nutritional value, taste, and prevents health risks from consuming stale fish. This study employs the Convolutional Neural Network (CNN) method with the VGG-16 architecture to classify fish freshness based on eye images. The dataset used consists of 1,903 fish eye images, augmented to 4,560 images. Classification results indicate that the VGG-16 model can distinguish between fresh and stale fish eyes with an accuracy of 85.26%. This research is expected to assist the fishing industry in monitoring fish quality more effectively and efficiently, as well as enhancing the safety of fish consumption for the community.