When CodeWizards presented their challenge of improving sustainability, we, as a team were sure to tackle it. Back home in India, we have heard of numerous cases of farmers committing suicide due to crop failure and crop loss. Alongside, the amount of resources (pesticides etc.) that are invested in growing crops that eventually fail, is a huge amount, both in terms of cost and the environmental impact they have. To tackle this issue, we wanted to use machine learning to help the government and farmers predict their yield at the end of each season depending on various factors such as area, season, weather conditions, Methane levels, soil quality etc.

What it does

Using Machine learning, we have trained the Random Forest Regressor model on the yield that was produced in previous years as a result of input parameters such as area and season. The model has been tested for accuracy and can be used to approximate the yield for future years. The model can be embedded into a web or mobile application but due to shortage of time we have not been able to embed it yet. The concept has been demonstrated through mobile application graphics.

We do wish to delve deeper than the top level prediction approach we have used. We do wish to let this be a personalised service for each farmer to let them monitor their farms, themselves. We aim to introduce more parameters into consideration, other than the existing parameters of 'area' and 'season', such as Soil Quality (pH sensor) etc. that would help the farmer detect any issues in their farm early on and accordingly invest resources (eg. pesticides, irrigation methods etc.) to fix avoidable issues.

How we built it

We built it using python, html, css and graphics.

Challenges we ran into

We got stuck at merging a ML code written in Python with an iOS Swift app.

Accomplishments that we're proud of

First Hackathon and a working code!

What we learned

Through the presence of the sponsors and experienced hackers, we learnt quite a few tricks to do things quicker!

What's next for Machine Learning to predict end-of-season Crop Yield

Delve deeper and make it a personalised service for each farmer to access and monitor their own farms!

We also wanted to address the 'Goose challenge' using our ML skills so we wish to develop a program that is trained on images of the AstonGoose. We then hope to add a scanner and goose sensor to detect goose on the farm. If and when detected: CHOP, CHOP, CHOP.

Built With

Share this project: