Crops+ | Devpost

Logo
results after the form is submitted
algorithm description for the users
home page of the website
working backend

Inspiration

I recently watched a documentary about deforestation in the Amazon rainforest. It stated that a leading cause of the issue was that farmers cut down many trees in order to make space for farmland. I thought to myself, "we should really be making the most of the land we already use for farming instead of using more." This gave me the idea for Crops+.

What it does

Crops+ is an easy-to-use website that helps farmers maximize their agricultural efficiency. Farmers have to randomly selected 10 soil particle samples from their plot of land and input the diameter of each into the form on the website. This inputted data is then run though a kNN machine learning algorithm trained on a manually-constructed dataset. The algorithm then returns the soil type and ideal crops for plantation to the farmer.

How we built it

The front-end of the website was created using Wix Velo and the Wix API was utilized to add custom JavaScript to the website. The back-end of the website was composed of the Python Sci-Kit Learn library, FastAPI with Docker and Azure. SKLearn was used to train the kNN ML algorithm that determines soil type based on the diameter of 10 soil particulate samples.

We did our research and found that there are 12 types of soils scientifically. Each of the soil types are suitable for some specific types of crops and vegetation. Our algorithm predicts the soil type, and we provide the user with recommendations on what crops to plant.

Challenges we ran into

Integrating the ML algorithm into the Wix Velo website was particularly challenging as none of the team members had worked with Wix Velo before. Nonetheless, we gave it our best shot and after hours of struggle and Stack Overflow browsing, the ML algorithm was integrated successfully. Plus, it was a little difficult to deploy the API with scikit-learn. Ultimately, we used docker to containerize the app instead of running it directly on a server.

Accomplishments that we're proud of

We're really proud to have created a website that has the potential to have a real impact on global issues like deforestation. We're also really happy to have familiarized ourselves with new software tools like Wix Velo and Azure.

What we learned

We learned a lot about farming and crops. During our research period, we learnt about the damaging effects certain vegetation can have on certain soil types, the inefficiencies of certain crops on certain types of soil, among other things. We also learned that varieties of soil can be classified by the average size of their particulate. In terms of the software development aspect, we familiarized and enhanced our ability to work with dev tools like Wix and SciKit-Learn.

What's next for Crops+

We're planning to develop the website further to include a sign-in option. Users can sign-up and store the soil types of their various plots of land all in one place. This will make it so that they'll be able to easily access ideal crops for each of their plots of land during plantation season, this will also allow features like sharing the results with partners or collaborators.