Krishi Unnati

Krishi Unnati is a mobile application for plant and crop disease detection using convolutional neural networks. We have used convolutional neural networks (CNNs) to build and train a model which can classify crop and plant diseases from the images of their leaves. Our model has achieved a test accuracy of 86.57% by using transfer learning on a pre-trained VGG16 model.

• You can find the initial Jupyter Notebook here. This contains the code for the initial model we built which had fewer classes of diseases and resulted in a test accuracy of 96.61%.

• You can find the Jupyter Notebook for an updated model for InOut 7.0 final submission here.

Team Members :

1. Prateek Senapati
2. Sudanshu Pandey
3. Sumit Kumar Mishra
4. Milind Pandey

Mobile Application:

Here are some screenshots of the Krishi Unnati mobile application:

1. Potato Late_blight

2. Corn__Common_Rust

3. Blueberry_healthy

4. Apple__Apple_Scab

5. Grape_Esca_(Black_Measles)

Note:

Krishi Unnati has been built on top of an existing project which was a submission for one of our academic courses. The project we had submitted for our academic course contained only the Jupyter Notebook which had the code for our initial model with a test accuracy of 96.61%. There was no mobile application back then. Our submission for InOut 7.0, Krishi Unnati, has substantial changes and additions. For our InOut 7.0 submission, we have built the whole model from the start with addition of new classes of diseases. Moreover, we have also developed a mobile application for Krishi Unnati.