SmallFishBD: An extensive image dataset of common native small fish species in Bangladesh for identification and classification

This data article presents a comprehensive image dataset of ten native small fish species commonly found in Bangladesh: Bele (<i>Glossogobius giuris</i>), Chanda Nama (<i>Chanda nama</i>), Chela (<i>Salmostoma bacaila</i>), Guchi (<i>Mastacembelus pancalus</i>), Kachki (<i>Corica soborna</i>), Mola (<i>Amblypharyngodon mola</i>), Kata Phasa (<i>Stolephorus tri</i>), Pabda (<i>Ompok pabda</i>), Puti (<i>Puntius sophore</i>), and Tengra (<i>Mystus vittatus</i>). The dataset was carefully curated to facilitate the study and research in fish species identification, classification, and biodiversity monitoring. Specimens of these species were collected from various fish markets in the capital city Dhaka. Different varieties of fish are supplied to Dhaka city from diverse geographical locations in Bangladesh. Thus, the dataset ensures a representative sampling of local aquatic biodiversity. To maintain uniformity across samples, images were captured using a smartphone camera under a standardized and controlled environment. Each specimen was placed against a neutral background with consistent lighting conditions. This limits environmental variability and enhances image quality for analytical use. The dataset contains high-resolution original images that were augmented using standard data augmentation techniques. This augmentation introduced variations such as rotations, flipping, and brightness adjustments. This expands the dataset and improves its utility for training robust machine learning (ML) and deep learning (DL) models in computer vision applications. The dataset has significant reuse potential across multiple domains. It serves as a critical resource for researchers and industry experts to develop automated systems for fish species identification and classification, particularly in the context of the rich aquatic biodiversity in Bangladesh. Furthermore, the dataset can facilitate ecological and environmental studies and research by supporting the monitoring of native fish species distribution and population dynamics. Its structured format facilitates integration into ML/DL pipelines that can foster advancements in fisheries management, sustainable aquaculture, conservation biology, and economic and cultural studies. Thus, the dataset represents a significant step towards integrating technological advancements and ecological sustainability. This article outlines the utility of the data, the dataset structure, the data collection methodology, and the applied augmentation processes to ensure transparency and reproducibility for future research endeavors.