How I Helped My Village Farmer Using Deep Learning!

1. Project idea, How it started?

Last month I completed my deep learning course as part of my PGD from IIIT Bangalore, I have done many projects based on CNN and RNN, LSTM but with the publicly available data. Even though those projects and assignments looks cool but that doesn’t gave me much satisfaction. so, I decided to do one project from scratch.

2. why this project?

I live in a village! so I choose an agriculture-based project and it also very easy for me to collect the data.To collect the data first I need to think of a problem right?
So I interacted with one of my nearest okra plants field farmer and I asked him several questions like what is the major disease that affects the growth of okra plants?
he’s not able to say the exact disease name but he showed me some diseased okra leaves and those leaves are completely affected by the disease(yellow vein mosaic) and he also mentioned that during an early stage of disease spread it is very difficult to spot the disease(as healthy and diseased almost looks similar) and by the time he identifies the disease it almost spread over remaining plants. so after walking around the field and talking to him, I found out the problem and that is detecting the disease at an early stage is a crucial thing and if we can able detect the disease at an early stage, we can stop the further spread of disease by using pesticides.
so finally I formulated the problem!🕺

3. what happened next?

After I found the problem that can be solved using deep learning, I discussed the same with my friend and he agreed to help me.
As the next step of the process, we collected the image data of okra diseased and healthy leaves within two days and we have around 2000 images ready for training.

Curious to look at data?😃

here are some sample images :

Yellow vein mosaic diseased leaf

Healthy okra leaf

4. The goal of this project?

The main goal of this project is to create a model that can detect the disease(yellow vein mosaic) and suggest pesticides to control the further spread of the disease.

You can find the notebook in my github profile :

manoj567/yellow-vein-mosaic-disease-prediction-in-okra-leaves

After model training, I created a web application using flask( for pesticide suggestion and prediction)

If anyone interested in improving the accuracy further, you can find the dataset here: https://www.kaggle.com/manojgadde/yellow-vein-mosaic-disease

Please feel free to connect with me on Linkedin :)

Manoj Gadde - Kaggle contributor - Kaggle | LinkedIn