BDFreshFish: A Comprehensive Image Dataset for Machine Learning Applications on Bangladeshi Freshwater Fishes

The "BDFreshFish" dataset contains a set of image data for eight different categories of native freshwater fish from various parts of Bangladesh. The eight distinct classes of fish have the following scientific names: Anabas Testudineus, Batasio Tengana, Channa Punctata, Marcrobrachium malcoimsonii, Heteropneustes Fossilis, Mastacembelus Armatus, Ompok Bimaculatus, and Puntius Sophore. Our collection comprises 3100+ images across these eight classes, with 300 augmented images and approximately 100 raw images in each class. The majority of raw images originate from rivers in Bangladesh. The dataset has four primary fish species characteristics: head, body, scales, and fins. Each original image was taken in adequate natural light against the right backdrop. Each image was properly organized into an appropriate folder, allowing different machine learning and deep learning models to make the best use of the images. Utilizing this enormous dataset and various traditional machine learning (TML) and deep learning (DL) techniques, researchers could make significant advancements in analyzing particular fish category classifications to analyze the freshwater environment and fish movement.

"BDFreshFish"数据集收录了源自孟加拉国各地的8类本土淡水鱼的图像数据。该数据集涵盖的8个鱼类类别的学名如下：Anabas Testudineus、Batasio Tengana、Channa Punctata、Marcrobrachium malcoimsonii、Heteropneustes Fossilis、Mastacembelus Armatus、Ompok Bimaculatus以及Puntius Sophore。本数据集的8个类别共计包含3100余幅图像，每个类别配有300幅增强图像与约100幅原始图像，其中绝大多数原始图像拍摄自孟加拉国境内的河流。

该数据集覆盖鱼类的4类核心形态特征：头部、躯干、鳞片与鱼鳍。所有原始图像均在充足自然光下搭配合适背景拍摄完成，且所有图像均已按规范整理至对应文件夹，可支持各类机器学习与深度学习模型充分利用该数据集开展研究。

研究人员可借助该大型数据集，结合各类传统机器学习（TML）与深度学习（DL）技术，在特定鱼类类别分类分析方向取得重要进展，进而开展淡水生态环境与鱼类活动的相关研究。