NativeAnimalBD: A Hierarchical Image Dataset of Bangladeshi Native Animals

The Bangladeshi Native Animal Image Dataset (NativeAnimalBD) is a curated collection of 8,000 high-quality images representing five major classes of native Bangladeshi fauna. The dataset is designed for research in computer vision, hierarchical image classification, and deep learning. The dataset is organized hierarchically, with five main classes, each comprising five subclasses. Every subclass contains 400 images, resulting in 2,000 images per main class. The hierarchical structure is as follows: Aves: Heron, Kingfisher, Magpie, Sparrow, Parrot Fish: Byen, Koi, Sing, Taki, Tengra (empty folders; images available via the BDINDIGENOUSFISH2019 dataset: GitHub link) Insecta: Ants, Bee, Butterfly, Dragonfly, Grasshopper Mammals: Asian Elephant, Indian Leopard, Royal Bengal Tiger, Spotted Deer, Wild Boar Reptiles: Crocodile, Gharial, Lizard, Snake, Turtle Dataset Structure:The dataset is arranged in a folder-based hierarchy, where each main class contains separate folders for its subclasses. This organization facilitates hierarchical classification experiments, benchmarking, and reproducible research. The dataset is distributed as separate compressed files for each main class (Aves, Fish, Insecta, Mammals, Reptiles) to ensure stable downloading and efficient file management. Intended Use: "Academic research in computer vision and machine learning" "Hierarchical animal classification studies" "Educational purposes and dataset benchmarking" License and Attribution: Images (excluding the Fish class) were collected from Unsplash and Pexels and remain subject to their respective licenses. Users must comply with the original license terms of these platforms. Fish images are not redistributed in this dataset. Users are advised to obtain them directly from the BDINDIGENOUSFISH2019 repository.