DETEKSI KUALITAS RUMPUT LAUT MENGGUNAKAN METODE CONVELUTION NEURAL NETWORK (CNN) BERDASARKAN CITRA DIGITAL (STUDI KASUS : DESA ALASMALANG KECAMATAN RAAS SUMENEP)

Supandi Supandi (Universitas Ibrahimy)
Abd. Ghofur (Universitas Ibrahimy)
Firman Santoso (Universitas Ibrahimy)

RABIT: Jurnal Teknologi dan Sistem Informasi Univrab
Vol 10 No 2 (2025): Juli

Seaweed quality greatly determines the economic value and competitiveness of processed marine products. To assist the automatic and objective quality classification process, a Convolutional Neural Network (CNN)-based approach is used with RGB color features from digital images as the main input. Seaweed images go through the stages of preprocessing, RGB color feature extraction, quality class labeling, and CNN model training. Two quality categories are defined, namely good and bad. The training results show that the CNN model is able to classify seaweed quality with an accuracy rate of 92% on training data and 91.19% on test data, as well as low and stable loss values. The application of CNN to color features has proven effective for image-based seaweed quality classification, and can be further developed in an automatic quality evaluation system in the agricultural and marine sectors.