Fighting fish identification using deep learning / Muhammad NurSyafiq and Mohammad Hafiz Ismail

• Main Authors: NurSyafiq, Muhammad, Ismail, Mohammad Hafiz
• Format: Book Section
• Language: English
• Published: College of Computing, Informatics and Media, UiTM Perlis 2023
• Subjects: Machine learning; Neural networks (Computer science)
• Online Access: https://ir.uitm.edu.my/id/eprint/100516/1/100516.pdf; https://ir.uitm.edu.my/id/eprint/100516/
• Institution: Universiti Teknologi Mara

Fighting Fish is an ornamental pet that has been known by many people. It is native to southeast Asia. This type of fish has around 72 recognized species. Mostly known by their brightly coloured fins and aggressive behaviour and it became favourites fish to keep as pet because of these features There has been increasing trend to collect ornamental pet each year. The project is about using a pre-trained Convolutional Neural Network (CNN) model to identify different type of fish such as Halfmoon, plakat, rose tail and crown tail. In this paper we focus on five species of Betta fish. We used a MobileNetV2 as our model to classify the fish and using python to implement model using deep learning library such as Keras and TensorFlow. We used NasNetMobile to compare our pre-trained model performance and accuracy. Lastly, we integrate the model to mobile application to make fish identification easier.