Growify

The Agriculture of Tomorrow. Growify centralizes farming data, recommends crops based on measurements, connects communities, and incentivizes usage through cryptocurrency.

Comment 2

💡 Inspiration

Every day, we read about obesity issues and the lack of incentives to begin farming due to a lack of room and expertise. We have no choice but to consume food from supermarkets, which is both costly and potentially genetically modified. Every year, over 20 billion pounds of food are wasted on farmland in the United States. A typical homeowner is reminded to grow fruits and vegetables in their backyards, but instead opts for fast food to prevent the time-consuming challenge of determining if the soil is clean. What if an object could use the data from soil moisture, humidity, and temperature sensors to score the atmosphere on a scale of one to one hundred percent? What if it could detect temperature, humidity, pH, and rainfall, and then predict the best crop to grow in those conditions? Is it possible to use deep learning picture classification to divide crops into various groups in order to increase the model's utility? Can we use incentives to encourage users to use our product while still encouraging them to love farming?

🌱 What is it?

Our solution focuses on digitizing farming, and combining maintenance with community. Whether on a large scale farm or an urban roof, our solution will prioritize data centralization, community/social building and live crop measurements.

🚧 How we built it:

  • Frontend: Built in React, with the use of SCSS, ChakraUI, TailwindCSS, and the Bootstrap and ChartJS plugins which were all deployed to Vercel.

  • Backend: The backend utilizes NodeJS, ExpressJS, GraphQL and Redux for authentication and serverless storage.

  • Machine Learning: Built with Python along with Tensorflow, OpenCV, SciKit-Learn, Pandas The three models that were created include a VGG19 + feed forward, a random forest regressor and a random forest regressor classifier.

  • Cryptocurrency: Python, Sha256 deployed to Repl.it

  • 3D Model: Sketchup Pro, A-Frame, 3.js, HTML deployed to Glitch

  • Branding: Figma, Adobe Creative Suite

  • Pitch Video: Davinci Resolve, Photoshop

⛔ Challenges we ran into:

  • Building a beautiful interface in just 24 hours without using Figma as a prototype
  • Building a functional Cryptocurrency with fluctuations in little time
  • Struggling through vaccine pain and carpal tunnel syndrome
  • Finding data for the machine learning models that was relevant and useful (some data had to be simulated)
  • Building a publicly deployed ML app that could access and use our models
  • Ensuring that each new block was validating the chain as it is very difficult to construct a function that checks whether the relation of each block (the hash code) corresponded to the block on the right.
  • Furthermore, working virtually was incredibly difficult. However, we had music blasting on our server and focused on collaborating from our desks.

🏁 What we learned:

  • Aiming for the moon is worth it, even if we land short, we would have hit the stars!
  • Aaryan: I learned how to use Sketchup Pro in just about 3 hours to create a model of our Product, helped create some of the 2D images using the Adobe Content Suite and Figma. Additionally, I was able to help my teammates touch up their designs.
  • Veer: Prior to this hackathon, I was great with machine learning and data science, I could create accurate models that make viable predictions. However, in this hackathon, I had to build my own dataset to simulate the Arduino sensor readings. I also learned how to use Streamlit to deploy the ML models and make predictions! I even experimented with cockroachDB and attempted to upload the csv of optimal soil measurements but could not get the end to end solution working. Nevertheless, I still learned a lot which can be applied in countless applications!
  • Rajan: I was confident in my frontend skills, but actually building the backend for the project was something I learned. I have barely used serverless with GraphQL or Redux, so this was an incredible learning experience.
  • Jaival: Although I was confident about my Python skills and my ability to create a cryptocurrency, building a live blockchain and coin was a little bit difficult. I had to learn how to manipulate the value of a coin based on transactions and also learn how to create a function that can validate the blockchain by checking the hashcode of each block everytime a new block is added. In terms of the Ar/Vr, I had some basic knowledge on how to use A-Frame. However, I had trouble importing .gltf files as they were not rendering on my editor, hence, I had to learn how to use and export 3D models as .obj files and how to use glitch (another script editor) to create the web-based Vr application.

📈 What next for the future of Growify:

  • Create a real life, prototype of our 3D Model
  • Launch a Kickstarter Campaign to test the market and check for demand verification
  • Use industrial sensors (read NPK ratios and pH levels)

Built With

+ 4 more
Share this project:

Updates