Convolutional Neural Network

Implemented a Convolutional Neural Network for the classification of tomato leaves diseases', having in total 11 classes and data for each class. There's data for training and validation for each one of the classes listed in the dataset.

SMA0401C

Uses a framework to implement a Deep Learning Model

The Convolutional Neural Network is implemented as a Deep Learning Model, using TensorFlow and the Keras API.

Evaluate the model with its initial configuration and make adjustments to improve its performance

The model is evaluated with a test and validation set, with an initial configuration and an improved version. Since the first model had overfitting, regularization techniques (dropout) was implemented.

Uses a real dataset and not one used for the class for the model development

The Tomato Leaf Disease dataset from Kaggle has real images used for the classification of diseases that can be found in tomato leaves.

File

Analysis and Report on Model Performance1

This is the file that contains the analysis on the model performance.

weights_1.h5

The weights from the first trained model. This file can be found in the link under the DATA section in the Analysis and Report on Model Performance1.

weights_2.h5

The weights from the second trained model. This file can be found in the link under the DATA section in the Analysis and Report on Model Performance1.

tomato1.zip

The dataset used for the models training process. This file can be found in the link under the DATA section in the Analysis and Report on Model Performance1.

cnn.ipynb

This is the file that needs to be executed to train the models and validate them.