Plant diseases detection

This project uses Neural Networks to detect and classify diseases in plants

Dataset

You can get the dataset from Kaggle

Followed steps

1. Use transfer learning using the imagenet weights
2. Add the last layers
   • Include the dropout layer for later optimization
3. Create the train and valid generators
4. Look for the best hyper parameters using the Bayesian Optimization
5. Fit the model
   • Using these hyperparameters
   • Including Early Stopping to avoid overfitting
6. Prune the model
   • Using the prune_low_magnitude method
   • Using the Polynomial Decay
7. Compress the model using the strip_pruning method
8. Evaluate the accuracy
9. Evaluate the metrics
   • Classification report
   • Confusion matrix
10. Test the model

Final results

Classification report

Confusion matrix

Some interesting results