MatsyaDx-BD: An image dataset of freshwater fish diseases from aquaculture farms in Bangladesh

MatsyaDx-BD is a curated image dataset of freshwater fish diseases collected from aquaculture farms in the Rajshahi Division, Bangladesh. The dataset contains a total of 2137 RGB images representing three widely farmed freshwater carp species: Grass Carp (Ctenopharyngodon idella), Rohu (Labeo rohita), and Silver Carp (Hypophthalmichthys molitrix). These images cover four major health conditions commonly observed in freshwater aquaculture systems, including Bacterial Gill Disease (275 images), Bacterial Red Disease (653 images), Epizootic Ulcerative Syndrome (EUS) (332 images), and Healthy Fish (877 images). The fish body length ranges are 20–44 cm for Grass Carp, 23–40 cm for Rohu, and 32–50 cm for Silver Carp. All images were captured using a Samsung Galaxy S25 Ultra smartphone under natural aquaculture farm conditions, resulting in realistic variations in lighting, background, and fish orientation.

In this updated version, several improvements have been introduced to enhance the dataset’s usability, structure, and reproducibility. The dataset has been reorganized into a specimen-wise folder structure consisting of 86 distinct specimens, which enables better traceability and supports fine-grained analysis at the individual specimen level. Additionally, the metadata has been improved and enriched to provide clearer annotations and facilitate easier integration into machine learning and computer vision workflows. All images have been uniformly resized to a resolution of 4000 × 3000 pixels to ensure consistency in input dimensions across different models and experiments. Importantly, no changes have been made to the original raw data content, thereby preserving the authenticity and integrity of the dataset.

The dataset is organized in a hierarchical structure where images are grouped by individual specimens and annotated with corresponding disease labels and species information. This enhanced organization supports both specimen-level analysis and traditional class-wise disease classification tasks, making the dataset suitable for applications such as deep learning-based fish disease detection, computer vision research in aquaculture, IoT-based smart aquaculture monitoring systems, and regional disease pattern analysis. The dataset is released under the Creative Commons Attribution 4.0 International (CC BY 4.0) license, allowing reuse and distribution with proper attribution.