FishSpeciesBD: A Real-World Fish-Market Image Dataset for Hilsa Authentication and Visually Similar Fish Species Recognition in Bangladesh

FishSpeciesBD is a real-world fish-market image dataset for Hilsa authentication and visually similar fish species recognition in Bangladesh. The curated dataset contains 18,957 RGB JPEG images representing five fish classes commonly encountered in Bangladeshi markets: Hilsa shad (Tenualosa ilisha), Toli shad (Tenualosa toli), Indian oil sardine (Sardinella longiceps), Goldstripe sardinella (Sardinella gibbosa), and Dotted gizzard shad (Konosirus punctatus). The curated image distribution is 14,318 Hilsa shad images, 1,073 Toli shad images, 2,243 Indian oil sardine images, 370 Goldstripe sardinella images, and 953 Dotted gizzard shad images. Curated images are stored as RGB JPEG files standardized to 512 × 512 pixels. The dataset also includes metadata, validation reports, checksums, documentation, scripts, and a supplementary original_images folder for transparency.FishSpeciesBD 数据集是一款面向孟加拉国鲥鱼真伪鉴别与视觉近似鱼类识别的真实渔市图像数据集。该经精心筛选整理的数据集包含18957张RGB JPEG格式图像，涵盖孟加拉国渔市中常见的5个鱼类类群：鲥鱼（Hilsa shad，Tenualosa ilisha）、托里西鲱（Toli shad，Tenualosa toli）、印度油沙丁鱼（Indian oil sardine，Sardinella longiceps）、金带沙丁鱼（Goldstripe sardinella，Sardinella gibbosa）以及斑腹𬶭（Dotted gizzard shad，Konosirus punctatus）。 各品类图像的样本量分布如下：鲥鱼14318张、托里西鲱1073张、印度油沙丁鱼2243张、金带沙丁鱼370张以及斑腹𬶭953张。所有精选图像均采用RGB JPEG格式存储，且统一标准化为512×512像素尺寸。为保证研究透明度，该数据集还附带元数据、验证报告、校验和、文档资料、运行脚本，以及用于溯源的original_images补充原始图像文件夹。