Prediction Model of Tuna Loin Quality Based on Image Processing in the Fisheries Processing Industry on Ambon Island

Indonesian fisheries waters hold significant fish resources, constituting about 37% of the world's total fish species. Among these, tuna is a key commodity, particularly in Maluku Province, which has experienced substantial growth in tuna production and exports. However, traditional methods of processing and assessing tuna quality, particularly for loin cuts, pose challenges due to inadequate facilities and manual sorting, which can lead to contamination and misclassification of fish quality. To address these issues, this study developed a deep learning model based on Convolutional Neural Networks (CNNs) to automatically classify tuna loin meat grades into two categories: Alpha and Bravo. The model was trained on a dataset of tuna loin images, achieving 100% accuracy. The model was subsequently integrated into a simple web application for real-time classification. Testing with new data demonstrated that the model performs well, accurately classifying the tuna loin into the correct grades. However, some misclassifications indicate the need for further refinement, particularly during data processing. Future research should focus on improving the model by increasing the size and diversity of the training data, refining data-processing techniques, and exploring the potential of mobile applications. This work represents a significant advancement in AI-based food quality assessment and has practical implications for the seafood industry, particularly in improving the efficiency and accuracy of tuna grading processes.