FreshWaterFishBD-28: A Dataset of 28 Freshwater Fish Species in Bangladesh

1.Freshwater fishes are an essential part of Bangladesh’s aquaculture and play a crucial role in nutrition, commerce, and local livelihoods. However, accurate identification of species is often challenging due to the diversity of fishes, their similar appearances, and variations in environmental conditions. This Freshwater Fish Dataset is carefully curated to assist researchers, aquaculture specialists, and machine learning practitioners in identifying and classifying freshwater fish species. This dataset contains high-resolution images of 28 fish species, collected from four different marketplaces in Gazipur under diverse environmental conditions.
2.The dataset includes the following species with the corresponding number of images:Aair (1804), Black Roi (306), Boal (1710), Chapila (428), Common Carp (517), DeshiPuti (412), Foli (562), Grass Carp (410), Ilish (1031), KalBaush (917), Katla (2196), Koi (842), Magur (574), Mirror Carp (415), Mrigel (2213), Nile Telapiya (409), Pabda (1764), Pangas (1394), Puti (1560), Rui (3240), Shing (60), Shol (1424), Shorputi (44), Silver Carp (400), Taki (2223), Tarabaim (1262), Telapiya (2058), and Tengra (1431).
3.Modern computer vision and deep learning approaches offer tremendous potential for automated fish species recognition and monitoring, which can enhance fisheries management, conservation efforts, and aquaculture operations.
4.With a total of 31,606 images, captured under various lighting, angles, and market settings, the dataset provides real-world variability for robust AI model development. Images can be further augmented using techniques like flipping, rotation, zooming, and brightness adjustment to expand the dataset and improve model performance. Species annotations were verified with local fisheries experts to ensure high-quality and reliable labeling.