SC008

FishSnapID: Fish Identification with Deep Learning for Malaysia Local Fish Market

Muhammad Faris Bin Shaik Mohamed, Muhammad Yazid Bin Yunos, Dr. Aziah Binti Ali, Dr. Noramiza Binti Hashim

AFFILIATION
Faculty of Computing & Informatics, Multimedia University

Description of Invention

The FishSnapID project uses deep learning to develop a mobile app for identifying fish species in Malaysia. Employing DenseNet121, VGG16, and MobileNetV2, DenseNet121 achieved 99.78% accuracy with a low test loss of 0.01%. The app lets users capture or upload images for identification and provides educational resources about fish species and habitats. This initiative addresses the challenges of limited local datasets.