Description : Here I used Artificial Intelligence in diagnosing plant diseases. Various diseases like early blight and late blight immensely influence the quality and quantity of the potatoes and manual interpretation of these leaf diseases is quite time-taking and cumbersome. Therefore I created a Web App using Streamlit which simply classify Potato Leaf Diseases and, finally deployed the Web-app on Heroku. Internally, our model is built using a simple Convolutional Neural Network Architecture to classify Potato Leaf Diseases. Initially I collected ready-made data from internet. Then due to small size of dataset, I used one of the simple and effective method, called Data Augmentation to increase the size of dataset as well as to reduce overfitting of our model. At the end built a Deep Learning Model to detect or classify Potato Leaf Diseases and got a test accuracy of 97%.
Heroku App : https://potato-leaf-disease-detection.herokuapp.com/
Dataset Source : https://www.kaggle.com/arjuntejaswi/plant-village
Article Link : https://www.analyticsvidhya.com/blog/2021/12/end-to-end-potato-leaf-disease-prediction-project-a-complete-guide/
Folder Structure :
Potato Leaf Dataset --> main folder
----| train
----| Potato_Healthy
----| img1.jpg
----| img2.jpg
----| img3.jpg
----| Potato_Early_Blight
----| img1.jpg
----| img2.jpg
----| img3.jpg
----| Potato_Late_Blight
----| img1.jpg
----| img2.jpg
----| img3.jpg
----| test
----| Potato_Healthy
----| img1.jpg
----| img2.jpg
----| img3.jpg
----| Potato_Early_Blight
----| img1.jpg
----| img2.jpg
----| img3.jpg
----| Potato_Late_Blight
----| img1.jpg
----| img2.jpg
----| img3.jpg
----| valid
----| Potato_Healthy
----| img1.jpg
----| img2.jpg
----| img3.jpg
----| Potato_Early_Blight
----| img1.jpg
----| img2.jpg
----| img3.jpg
----| Potato_Late_Blight
----| img1.jpg
----| img2.jpg
----| img3.jpg
Sample Output : The output is showing 3 thing's.
- Predicted Class : The model's output.
- Actual Class : The actual output.
- Confidence : How confident our model is.