School of Graduate Studies, Mindanao State University – General Santos City 9500, South Cotabato, Philippines
Corresponding author
malonava@pnri.dost.gov.ph
The Philippines is the second biggest source of skipjack tuna (Katsuwonus pelamis), contributing to the country’s economic development. However, its sustainability faces challenges due to overfishing and a lack of proper management practices. Otoliths are important tools for managing fish stocks, but their analysis is time-consuming and requires a high level of expertise. In this paper, we explored the use of convolutional neural networks (CNNs) to recognize patterns and classify them according to developmental stages. The results showed that the CNN model achieved an accuracy of 100% in classifying otoliths by developmental stage using the RMSprop optimizer, demonstrating the potential of deep learning to provide a standardized and reliable protocol for managing fish stocks in countries like the Philippines, where there is a shortage of trained fish experts. This study provides an innovative approach to guide future efforts in conserving fish populations and promoting sustainable fishing practices.