Group 55 — ReLeaf

"Giving your sense of relief, everywhere": An intuitive app that helps users in their meditation journey using machine learning and a simple, clean, UI.

Comment

Inspiration

When was the last time you sat before an exam or interview and told yourself, “Wow, I’m gonna crush this, and if I don’t, that’s okay”? The inspiration for the project came from one of the members, who vividly remembers their hands shaking while writing a test and heart pounding before an interview. They wanted to create an app that prevents or lessens the anxieties associated with everyday stresses, like exams, interviews, etc., and provides a sense of relief.

While anxious, going on the internet and searching up calming exercises may cause overstimulation by the abundance of resources, an individual may become overwhelmed by choice. ReLeaf is the perfect method to get customized exercises and resources for your tailored situation, including meditation, tips on relaxing, a camera that helps you become more confident using machine-learning, and more.

ReLeaf will allow individuals to calm their nerves from everyday stresses and provide suggestions when general anxiety-inducing events are on the horizon. We wanted to develop something that the majority of individuals will benefit from, decrease anxiety and increase confidence, and help individuals overcome obstacles.

What it does

ReLeaf is a wellness app that provides an immediate guided meditation when a user feels anxious, or an exercise that is tailored to the user based on what they generally get anxious about. For example, 93% of Americans have experienced anxiety related to their interviews. If the user selected interviews as a common stressor, they can perform pre-interview activities that will ease their anxieties and build confidence. Some exercises include listening to their favourite song, calling their best friend, and deep breathing. Additionally, ReLeaf has built-in resources like motivational quotes and a camera that detects facial expression and body language, which uses machine learning and encourages you to pose your body to feel more confident.

There are other common stressors like exams and social media that have specific suggestions or exercises to perform, all of which their goal is to relieve stress.

How we built it

We used Figma for initial UI/UX design, Bootstrap, HTML, CSS, and JavaScript for front-end development, Flask and TensorFlow for back-end development, and OpenCV.

Challenges we ran into

  • Most of our project comprised of developing the machine learning algorithm. It was the most daunting process where we had to continuously train the model over Google Colab and Tensorflow. The next step was to interface it with a device’s webcam and be practically compatible with every device–especially in mobile applications. For demonstration purposes, we set up a web app using Bootstrap that is very flexible to use–this environment can help us adapt our project for mobile uses in the near future.
  • When trying to find a cloud service to host our site, we found it too large for Heroku, and Azure was having licensing issues leading us to use Google Cloud as our hosting service.
  • We were only able to achieve 66% accuracy with our machine learning model do to a lack of available datasets. The dataset we used, FER2013 has never been able to reach an accuracy of 71.4% even in research. This is due to certain emotions made up disproportionately small portions sections of the dataset, i.e., disgust only included 436 of the 28709 training images.

    • Accomplishments that we're proud of

  • Navigating Liyi’s first Hackathon!
  • It was the very first hackathon for some of our team members and it was a great experience learning how to identify each other’s strengths and weaknesses.
  • We are especially proud of our UI/UX, creating sleek and intuitive designs for ReLeaf. Our emphasis on ReLeaf is that it is accessible so we put a lot of time brainstorming design choices and control flow.
  • We felt even prouder about tackling mental health roadblocks in a span of 24 hours, developing a better understanding of the stressors in youths’ lives as well as our own lives.

    • What we learned

  • How to use Flask and integrate Python into frontend
  • How to collectively map out our ideas onto Figma to create an intuitive web app design
  • How to train a machine learning model and export the model to be used in an application

    • What's next for ReLeaf

    Our next step is to create a mobile app, allowing for a more on-the-go experience. We would also like to expand our dataset for our emotion recognition model, enhancing the app's ability to recognize and respond to users emotion.

    To make this the ultimate and personal meditation app, we aim to improve this product by adding more in-depth user authentication in the near future through tools such as Firebase or Google Cloud. This aids in the customization of users’ profiles for a highly tailored experience.

    We want to add more scenarios, like social anxiety, family anxiety, and more tips, like grabbing a snack, words of affirmation, etc. Some of these ideas include:

    • Improving UI and UX of our design
    • Improve our 1-minute quick meditation timer
    • Add user customization
    • Collaborate with researchers to help improve our product and address key mental health issues our app could tackle

    Built With

    Share this project:

    Updates