FishNet: A High-Resolution Image Dataset for Automated Fish Species Recognition

Dataset Overview
This dataset comprises 2,455 high-resolution images of commonly found freshwater and brackishwater fish species. The images were captured under natural lighting conditions and are organized into eight folders, each based on a specific species.

Class Distribution
Mola Carplet (Mola): 405
Swamp Barb (Puti): 495
Mystus Catfish (Tengra): 431
Stinging Catfish (Shing): 549
Prawn: 225
Shrimp: 122
Dwarf Gourami: 228

Data Processing:
Original HEIF images were converted to JPEG, resized to 640×640 pixels, and renamed using Python scripts (convert_heif_to_jpg.py, resize_and_rename.py) provided in the code/ folder for reproducibility.

Folder Structure & Naming:
Images are organized by species folder, with filenames as _.jpg (e.g., Mola fish_1.jpg).

Purpose
This dataset supports the development of automated fish species classification systems using machine learning and computer vision. It is useful for applications in aquaculture, seafood identification, food quality control, and biodiversity monitoring. It also serves as a valuable resource for training and evaluating image-based recognition models in both research and industry contexts.