Fish-gres Dataset for Fish Species Classification

Published: 25 November 2021| Version 2 | DOI: 10.17632/76cr3wfhff.2


Fish-gres dataset consists of 8 fish species, where the number of images on each species varies from 240 images to 577 images. This difference is due to the availability of random samples taken in traditional markets in Gresik, East Java, Indonesia. All of the species can easily be found in the traditional markets where the dataset was created. Image acquisition was carried out under various background conditions and the number of fish species varies in the traditional market in Gresik Regency, East Java, Indonesia. The resolution of the acquisition image is 4160x3120 pixels, then we resize the image to 390x520 pixels. The species of the fish are as follows, Chanos Chanos (500 images), Johnius Trachycephalus (240 images), Nibea Albiflora (252 images), Rastrelliger Faughni (544 images), Upeneus Moluccensis (577 images), Eleutheronema Tetradactylum (240 images), Oreochromis Mossambicus (331 images), and Oreochromis Niloticus (564 images), respectively. For use this dataset, please cite this related article: Prasetyo, E., Suciati, N., & Fatichah, C. (2021). Multi-level residual network VGGNet for fish species classification. Journal of King Saud University - Computer and Information Sciences.



Institut Teknologi Sepuluh Nopember, Universitas Bhayangkara Surabaya


Computer Science Applications, Aquaculture