<![CDATA[Indonesia is home to a vast array of biodiversity, including various species of little tuna. However, the process of identifying little tuna species is still challenging due to their diversity. The Indonesian Society and Fisheries Foundation (MDPI), which has the task of collecting fisheries data manually, is prone to significant identification errors. Therefore, the author proposes the utilization of Deep Learning, a Machine Learning method due to its ability to model various complex data such as images or pictures and sounds. This approach can facilitate the identification process of little tuna. In this research, the Resnet-50 architecture is utilised in the modelling process with the original dataset of 500 images. In this study, several test scenarios were also applied. The best results obtained are global accuracy of 91% and matrix accuracy value of 95%. These results were obtained using an augmented dataset with some parameter adjustments to the model built. With these good accurate identification, the MDPI Foundation is expected to better manage fisheries data and use it to support sustainable fisheries management.]]>
Copyrights © 2025
