<![CDATA[Fish is one of the ingredients for consumption that provides high-quality protein and can help form a healthy lifestyle. The perishable quality of fish meat demands that consumers be smart in sorting out the fish to be consumed. Therefore, knowledge about the freshness condition of fish meat is important for consumers. This study tries to build a mobile-based application that applies the Deep Learning model, using architecture You Look Only Once (YOLOv4) and YOLOv4-Tiny to detect the freshness level of fish based on the eyes and skin of the fish. The level of freshness used is fresh, medium, and spoil. The dataset used by the model are images of Deho tuna fish (Euthynnus Affinis), Manglah fish (Priacanthus Tayenus), Solok fish (Rastrelliger Brachysoma), Mackerel fish (Scomber Australasicus), Kuwe Lilin fish (Caranx Melanophygus), Teribang fish (Nemipterus virgatus), Banyar fish (Restrelliger Kanagurta), and Kolong fish (Atule mate). The mAP achieved by YOLOv4 is 99.17% and YOLOv4-Tiny is 97.25%. The fastest processing time, the freshness level of fish meat in the application reaches 2.5 seconds per image for YOLOv4 and 0.15 seconds for YOLOv4-Tiny.]]>
Copyrights © 2023
