In this dataset we are provided with images that belong to 4 classes : diseased leaf , diseased plant , fresh leaf and fresh plant. The objective of this study is to create a CNN model to help us predict whether these image of the leaf/plant belong to the diseased category or the healthy category. Working with images is pretty memory consuming, especially if you read and preprocess all of them at the same time. The following approach avoids this problem in Keras, leaving more space in memory to use augmentation and/or loading pre-trained models. I have tried my best to comment all the important steps but if any step is not well explained please feel free to contact me.
sailyshah/cotton-disease-prediction
In this dataset we are provided with images that belong to 4 classes : diseased leaf , diseased plant , fresh leaf and fresh plant. The objective of this study is to create a CNN model to help us predict whether these image of the leaf/plant belong to the diseased category or the healthy category.
Jupyter Notebook