Bloomlight

• 

  Journaling with AI

• 

  Main Garden Screen with some Flowers

• 

  Flower Inventory

Inspiration

In our current generation, mental health is rampaging through many populations and individuals often forget to slow down, take a breather or even simply talking about their day. This can slowly build bad habits and develop into a large downwards spiral. We wanted to completely eliminate and promote healthy habit building! No one realizes how impactful doing simple tasks such as journaling about your day can do onto your mental health. We're all on our phones anyways, at the very least, let us make yourself feel better!

What it does

Bloomlight is a user-interactive app that involves a garden and a magical wise capybara named Captain Felix the Great! Here, you nurture your garden alongside with your mental health. Journaling in the journaling option allows for users to gain sunny points. Using AI, your journal is interpretted and you are rewarded for doing small daily tasks that our trained AI deems worthy of sunny points; such as small things like making your bed, excercising, drinking water. Gaining sunny points causes the sun bar to fill up weekly to keep their garden and cheerful, inviting critters and causing your flowers to blossom even more! Sunny points can be used to purhcase more items to customize your garden, adding flowers and critters. Failure to consistently fill the sun bar does not harm the user's garden but causes the garden to fall asleep and look dull, causing no interactions and a gloomy setting. Alongside the garden, there is a magical capybara who is always available to be a shoulder to lean on. This chatbot was trained with empathetic and understanding conversations in order to support and fully encourage users.

How we built it

We used Godot, a game development tool in order to build the front end, used ChatBase to build Felix the capybara, and we used facehugger sentiment model and further trained it with specific reinforcements to better understand daily tasks to reward. We designed models and icons for the app and animated them in Godot.

Challenges we ran into

Neither of us had made a mobile app before therefore, we did not know how to use Godot or any other game development tool. We also had never made our own sprites, so we had to find an app to draw the icons and models. Fine tuning the Face Hugger app was also extremely difficult as using the Face Hugger provided datasets, they were extremely large. Sizing down multiple times in order to get a correctly fine tuned model was difficult. Our computers were also a limiting factor, having weak GPUs and CPUs did not help in our processing time so we had to opt to use Google CoLab which was another learning experience.

Accomplishments that we're proud of

Learning so much in such a short time was definitely a challenge. We're proud of not giving up and seeing it through to the end, and giving our best shot at this new project.

What we learned

We learned how to digitally design an app, train a personality based chat bot, integrate Flask into a program. We also learned how to work in an extreme time crunch, this being our first Hackathon, we managed to somehow pace ourselves decently for this large, ambitious project between two people who have never touched this kind of thing before.

What's next for Bloomlight

Hopefully this isn't the end of this project! We want to continue developing this app cause we had so much fun programming and designing the app. There's so much we wanted to add to this project that we barely scraped the surface of.

Built With

• chatbase
• colab
• flask
• godot
• huggingface
• kaggle
• python
• pytorch
• tensorflow