Related Experiment Video
Updated: Dec 27, 2025
10:56
Long-term Behavioral Tracking of Freely Swimming Weakly Electric Fish
Published on: March 6, 2014
12.9K
Automatic Fish Population Counting by Machine Vision and a Hybrid Deep Neural Network Model.
Song Zhang1,2,3,4, Xinting Yang2,3,4, Yizhong Wang1
1College of Electronic Information and Automation, Tianjin University of Science and Technology, Tianjin 300222, China.
Animals : an Open Access Journal From MDPI
|February 28, 2020
Summary
This study introduces an automated fish counting method using a hybrid neural network for intensive aquaculture. The novel approach achieves high accuracy, offering a non-invasive solution for real-time population monitoring.
Area of Science:
- Aquaculture technology
- Computer vision
- Machine learning
Background:
- Intensive aquaculture requires accurate fish population data for intelligent management.
- Traditional fish counting methods are labor-intensive, time-consuming, and stressful for fish.
Purpose of the Study:
- To develop an automatic, real-time, accurate, and non-invasive fish counting method for offshore salmon mariculture.
- To improve upon existing fish counting techniques using advanced neural networks.
Main Methods:
- A hybrid neural network model combining a multi-column convolution neural network (MCNN) and a dilated convolution neural network (DCNN).
- MCNN captures features from various receptive fields using different kernel sizes.
- DCNN minimizes spatial information loss during network processing.
Main Results:
- The hybrid model achieved a counting accuracy of 95.06%.
- A Pearson correlation coefficient of 0.99 was observed between estimated and actual fish counts.
- The proposed method demonstrated superior performance compared to CNN- and MCNN-based approaches.
Conclusions:
- The developed hybrid neural network offers an effective solution for automated fish population counting in aquaculture.
- This technology can provide crucial data for optimizing feeding and breeding operations.
- The method enables objective, real-time, and lossless fish population assessment.
