Inspiration

The need to address climate change is as urgent as ever. While there are organizations that offer the opportunity to convert donations into planted trees, many newly seeded plants are at risk of failure due to a lack of funding or improper growing conditions. Donors may not even be able to see the impact that their contribution has. Additionally, not everyone has the logistical or financial means to actively take part in nurturing a plant, even if they wish to do so.

What it does

With iPlant.AI, we seek to remove the boundaries that prevent individuals from partaking in one of the impactful ways of contributing to a greener planet and to instill inspiration for a brighter future.

There are two types of users on iPlant.AI: those who are seeking to grow a plant (activists), and those who are growing the plants (growers).

The role of the activist is to help the environment by boosting a plant and promoting the growth of plants. The activist can use the platform to boost a plant, and that money is given to the grower to purchase resources to keep on growing the plants. The activist is essentially owning a virtual plant in which they know where the plant is located, the health of the plant, the grower information, and images to track the progress of the plant.

As a grower, the user can upload photos of plants they grow that can either be for private use to monitor plant health, or be public for other activists and growers to explore. If the plant is public and another user takes interest in the plant, the grower can earn income for nurturing the plant from the crowdfund.

We use AI to examine every plant in our platform and automatically determine the name of the plant, health status (whether it’s healthy, or has bacteria, fungus, virus, etc.) and give some insights on the health status. This way we can be as transparent about how the plant is

With our solution, everyone benefits. Growers can earn income and contribute by growing for others, meanwhile, those who are unable to invest the time to nurture their own plants can still be fulfilled through the updated progress of the plants from the growers.

How we built it

Front End: We designed the UI using Figma to create a mockup on how the product should look like. To implement the design, we used HTML for structure, Bootstrap, and CSS/Sass for custom styling on top of that. We used the popular framework React.js to add functionality and logic onto the web application as well as client-side routing. When arriving on the page, you are greeted with a stunning landing page which gives the user options on page navigation depending on what the user intends to do. With two roles in mind, the user can go the grower route or the activist route. That being said, the user has access to either as well.

If the user is a grower, they will immediately be taken to what we call the “Grower Dashboard” in which gives the grower access to important data such as plants posted, the ability to add a new plant to the collection, and current balance other users have donated for their hard work. The grower also has the ability to sell a plant. The user has to follow a set of easy instructions that includes sharing their location where the plant is located, upload a picture and you’re set! Our application makes it that much more simple!

If the user is an activist, they will be taken to the trending plants page where they have the ability to see all different kinds of plants growers from around the world grow on our platform. You can learn more about them, find out what species of plant it is, where it’s located and you can event reward the grower for doing an excellent job on growing the plant! You’re also given the option to grow a new plant in which you help a grower with the resources to grow a plant of and you get to virtually own it! You can keep track of it and aid the grower so you make sure it stays nice and healthy all while promoting a better future for the next generation. You can also check the health status of any plant you may own using AI and it can recognize any plant you own and just by looking at a picture it can tell you if the plant is healthy.

Figma mock: Figma Design

Back End:

The back end was written in Python using Flask. For data persistence, we used Firebase’s Cloud Firestore Database to store our plant and grower models. We also stored the images we planned to use in Firebase Cloud Storage.

Dataset: The dataset we used is from Kaggle. It consists of leaves labelled with plants and their status (i.e, healthy or have a disease). After cleaning our dataset we are classifying: Apple, Cherry, Corn, Grape, Peach, Bell Pepper, Potato, Strawberry & Tomato, and their health status.

Data Science: We designed, built and trained a Convolutional Neural Network from our dataset that takes images and outputs the type of plant and the health status.

Model Performance and loss Graph: Model Performance and loss

Challenges we ran into

We faced many obstacles throughout the weekend.

  • Collaborating virtually is challenging
  • Deep learning training takes a lot of computing power and time
  • Availability of datasets
  • Connecting our front-end with back-end (and deep-learning)

Accomplishments that we're proud of

Apart from completing this hack, we persevered through each challenge as a team and succeeded in what we put ourselves up to. Our team was successfully able to learn about neural networks and make accurate models in a very short amount of time. We were able to achieve high accuracy models (90% test accuracy) with high precision, recall and f1-score. This shows that our model has fewer false positives and negatives. Beyond that, we were able to make a web application that allows plant growers to check how their plants are doing and raise funds for it.

What we learned

  • Working as a team
  • Training a deep learning model (Convolutional Neural Networks)
  • Configuration management
  • Working with Firebase

What's next for iPlant.AI

We hope to first develop an Android App and target people in developing countries for this service. We want to enhance our models by identifying more classifications with higher accuracy, recall and precision. We aim to make our service more transparent by using AI to track the progress of plants, and we have a long-term plan to tokenize our plants as NFT’s.

Built With

Share this project:

Updates