APA 6th Edition Štifanić, D., Musulin, J., Car, Z. i Čep, R. (2020). Use of Convolutional Neural Network for Fish Species Classification. Pomorski zbornik, 59 (1), 131-142. Preuzeto s https://hrcak.srce.hr/249270
MLA 8th Edition Štifanić, Daniel, et al. "Use of Convolutional Neural Network for Fish Species Classification." Pomorski zbornik, vol. 59, br. 1, 2020, str. 131-142. https://hrcak.srce.hr/249270. Citirano 08.12.2021.
Chicago 17th Edition Štifanić, Daniel, Jelena Musulin, Zlatan Car i Robert Čep. "Use of Convolutional Neural Network for Fish Species Classification." Pomorski zbornik 59, br. 1 (2020): 131-142. https://hrcak.srce.hr/249270
Harvard Štifanić, D., et al. (2020). 'Use of Convolutional Neural Network for Fish Species Classification', Pomorski zbornik, 59(1), str. 131-142. Preuzeto s: https://hrcak.srce.hr/249270 (Datum pristupa: 08.12.2021.)
Vancouver Štifanić D, Musulin J, Car Z, Čep R. Use of Convolutional Neural Network for Fish Species Classification. Pomorski zbornik [Internet]. 2020 [pristupljeno 08.12.2021.];59(1):131-142. Dostupno na: https://hrcak.srce.hr/249270
IEEE D. Štifanić, J. Musulin, Z. Car i R. Čep, "Use of Convolutional Neural Network for Fish Species Classification", Pomorski zbornik, vol.59, br. 1, str. 131-142, 2020. [Online]. Dostupno na: https://hrcak.srce.hr/249270. [Citirano: 08.12.2021.]
Sažetak Fish population monitoring systems based on underwater video recording are becoming more popular nowadays, however, manual processing and analysis of such data can be time-consuming. Therefore, by utilizing machine learning algorithms, the data can be processed more efficiently. In this research, authors investigate the possibility of convolutional neural network (CNN) implementation for fish species classification. The dataset used in this research consists of four fish species (Plectroglyphidodon dickii, Chromis chrysura, Amphiprion clarkii, and Chaetodon lunulatus), which gives a total of 12859 fish images. For the aforementioned classification algorithm, different combinations of hyperparameters were examined as well as the impact of different activation functions on the classification performance. As a result, the best CNN classification performance was achieved when Identity activation function is applied to hidden layers, RMSprop is used as a solver with a learning rate of 0.001, and a learning rate decay of 1e-5. Accordingly, the proposed CNN model is capable of performing high-quality fish species classifications.