Deep learning based optimal fish species identification to maximize production in fish ponds.

Area of Science:

• Aquaculture and Fisheries Science
• Artificial Intelligence in Agriculture
• Environmental Monitoring

Background:

• Aquaculture is vital to India's economy, ranking second globally in fish production.
• Identifying suitable fish species for specific pond conditions is crucial for maximizing yield.
• Traditional methods lack efficiency in selecting optimal species for aquaculture.

Purpose of the Study:

• To develop a Deep Neural Network (DNN) model for identifying the best fish species for aquaculture.
• To utilize water quality parameters (pH, temperature, turbidity) for species selection.
• To enhance fish production through data-driven species recommendation.

Main Methods:

• A dataset of 196 samples was collected from ponds in Warangal, Telangana, India.
• Real-time water parameters (pH, temperature, turbidity) were measured using sensors.
• A Deep Neural Network (DNN) model was implemented and compared against Decision Tree, Random Forest, SVC, KNN, and Naive Bayes classifiers.

Main Results:

• The proposed DNN model demonstrated superior performance compared to other machine learning models.
• The DNN model achieved 100% accuracy in both training and testing phases.
• Accurate identification of fish species based on water parameters was achieved.

Conclusions:

• Deep Neural Networks (DNN) offer a highly accurate solution for selecting optimal fish species in aquaculture.
• The model's 100% accuracy highlights its potential to significantly boost fish production in India.
• This AI-driven approach can revolutionize aquaculture management by providing precise species recommendations.