Recommendation System for Farming :: Iot and ML

This project was coded by the team “Harvestify” as a part of the Indo Data Week — Hackathon for Good 2020.

We all know that agriculture is the backbone of our country. But do we know what’s at stake if we don’t act upon the agricultural practices that everyone is following right now?

Maybe we do. Maybe we don’t. Either way my team and I decided to do something about it.

Problem:

The old traditional methods aren’t now suitable to the soil and the weather conditions, that are changing for the bad. In-fact they will have a negative impact on the soil.

Sadly, even with the experience a farmer has, it is still not enough to judge about how the climatic conditions or the soil is going to affect the crop.

Why? Because the NRT data of the soil is very low, if not zero.

So we came up with a recommendation system for farmers that’ll help them in making Informed Decisions.

What we do?

1. Recommend the Crop to be planted.
2. The fertilizer to be used based on the condition of the soil.
3. Yield Predictions based on the season and geographical locations
4. Real time data of what’s happening inside the soil.
5. Automatic Irrigation System

While it was a handful of modules for a farmer’s package, we managed to pull it off. ( in flying colors )

What we built is a small PoC.

1. We have a sender and a receiver connected via LoRa ( state-of-the-art IoT tech )
2. The sender collects all the real time values of the soil in which it is connected, creates a pool of the data and sends it to a centralized repository of data.
3. Using the real time values and weather parameters and other meteorological factors, we recommend the crop to be planted.
4. Now based on the crop that was recommended, we have curated an algorithm that will give recommendations about how care is to be taken of the soil ( in an organic way ).
5. This is where it gets fun. A Dashboard to see the NRT data of what’s happening in the field, inside the soil. The soil moisture levels, various gases present over there and the temperature.
6. And an interactive Dashboard for the farmer to know about the yield predictions.
7. All of these integrated to the Google Assistant to give data right to the farmer’s finger tips.
8. The automatic irrigation system takes care of the water levels w/o the intervention of the farmer.

IoT Architecture

Now when a group of farmers get the sensors installed on their farms, all the senders send the data to a centralized repository. Which would be right open for awesome analysis and recommendations.

Possible Business Model:

1. Subscription based model for the farmers.
2. Selling invaluable data from the central repository to various government bodies for better decision making regarding the crops and soil.

Sender Module

Over the years of repeated analytics and recommendations, we can create a more richer soil, lesser carbon emissions and healthier food and most importantly food for everyone, all contributing to the SDGs.

While there were many more features we had in mind to add to this project, we were limited by time. NDVIs and pest predictions could give the farmer a deeper insight on when to harvest and to prevent a pest attack.

Find the entire code here!

Here’s our team:

N Nithin Srivatsav

Pareekshith US Katti

Shivashankar S

Until Next time!!