Related Experiment Video
Updated: May 23, 2025
04:23
A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
Published on: April 21, 2023
1.7K
SwinFishNet: A Swin Transformer-based approach for automatic fish species classification using transfer learning.
1Department of Electrical and Electronics Engineering, Faculty of Engineering and Architecture, Recep Tayyip Erdogan University, Rize, Turkey.
Plos One
|May 20, 2025
Summary
This study introduces SwinFishNet, an AI model for automatic fish species classification using Swin Transformer. It achieves high accuracy, enhancing seafood sustainability and market efficiency.
Area of Science:
- Computer Vision
- Artificial Intelligence
- Marine Biology
Background:
- Accurate fish species classification (FSC) is vital for sustainable fisheries, food safety, and market efficiency.
- Current methods may lack the precision needed for the complexities of the global seafood trade.
Purpose of the Study:
- To develop an automated fish species classification system using advanced deep learning.
- To leverage the Swin Transformer architecture for enhanced image recognition in fish species identification.
Main Methods:
- Utilized transfer learning with the Swin Transformer (ST) model, named SwinFishNet.
- Trained and evaluated on three diverse datasets: BD-Freshwater-Fish (12 classes), SmallFishBD (10 classes), and FishSpecies (20 classes).
- Applied image preprocessing and optimization using the AdamW algorithm, evaluating performance with metrics like classification accuracy and F1-score.
Main Results:
- Achieved high classification accuracies: 0.9847 (BD-Freshwater-Fish), 0.9964 (SmallFishBD), and 0.9932 (FishSpecies).
- Demonstrated the model's effectiveness in distinguishing between various fish species from images.
Conclusions:
- SwinFishNet offers a robust and accurate solution for automated fish species classification.
- The AI-driven approach significantly contributes to improving sustainability, food safety, and market efficiency in the seafood industry.
