A Small Sample Recognition Model for Poisonous and Edible Mushrooms based on Graph Convolutional Neural Network

Area of Science:

• Agricultural Science
• Computer Science
• Machine Learning

Background:

• Accurate identification of poisonous and edible mushroom crops is crucial for improving agricultural yield and quality.
• Existing methods may lack efficiency or accuracy in disease identification.

Purpose of the Study:

• To develop and evaluate a graph convolutional network (GCN) model for automatic identification of poisonous and edible mushroom crop diseases.
• To enhance the representation and generalization capabilities of disease identification models.

Main Methods:

• Constructed six GCN models of varying depths for poisonous crop and edible fungi identification.
• Utilized a pre-trained model from the PlantVillage dataset, followed by parameter adjustment and fine-tuning.
• Employed multi-neural network model synthesis through averaging and weighting of predictions.

Main Results:

• The GCN model integrating multi-scale category relationships and dense links demonstrated improved performance.
• Dense connection technology enhanced the model's representation and generalization abilities.
• Achieved an average recognition rate of approximately 91%, with accuracy increases of 1%-10%.

Conclusions:

• The developed GCN model effectively identifies diseases in poisonous and edible mushroom crops.
• The integration of multi-scale relationships and dense links significantly boosts model accuracy and generalization.
• This approach offers a promising solution for automated crop disease identification, improving agricultural outcomes.