2/Day

Inspiration

Due to COVID-19 and other current issues in the world, the increase of screen time for many had also meant more social media screen time, and there had been an unprecedented increase in depression and anxiety seen in the generation. It is becoming apparent that social media had brought many negative effects for this generation: Unrealistic expectations about body image and sources of self evaluation, normalization of risk taking behaviors, are all detrimental to people’s mental health.

Through current viral social media “BeReal” we got inspiration for the possibility of “Alternative Social Media”. Upon hearing about a machine learning model for generating sentences based on contents in a photo, we got the idea that we could leave all of the “presentation” work for AI to do and focus on showing the photos that truly brings joy and documents our lives rather than “packaging” ourselves on the internet. We got many inspiration to gamify and avoid any commentary/quality measurement of how good a person’s posing is through mechanisms in games like Splatoon and Arknights.

What it does

2/Day (2 per day) is a machine learning based app that serves as a photo sharing focused social media. It aims to build a fun and inspiring platform with healthy social media habits through taking 2 images (no retaking chance) each day with AI generated description. It works based on the machine learning model trained with input (images) and output with descriptive text sequences related to the images. In addition to that, we gamified the experience with “Word Bank” that user can build up over time through making posts, reviewing posts by friends, or published photos on Playground.

The special thing about the 2/Day app is that there are no means of measurement for how “good” a person’s photo is. The only thing that gives users a sense of what is a “good photo” is that popular photos most gets more chances to be shown to other people’s Playground.

With the experience of 2/Day, we want people to capture the purest memories they treasure and leave all the “presentation work” behind, and to get to know our true self more.

How we built it

We mainly used Figma for this project as the idea for the AI would require time for training and reviewing.

Challenges we ran into

We couldn’t really implement the AI in code since training and reviewing time was required.

Accomplishments that we're proud of

In our group of 3 people, only 1 person has attended a hackathon previously online. It is definitely a very rewarding and wonderful experience for everyone and just the experience in attending and finishing our project is an accomplishment we are proud of.

What we learned

Going into this project, we had only heard of AI technology before and didn’t really understand how it worked and could be used. Upon brainstorming through this idea, we learned much more about how AIs comprehend photos and process the image data into vectors and later into words. We also learned the startling truth in statistics about current social media influences through research. It made us more determined to find a better solution for ways people share online.

What's next for 2/Day

There were definitely a lot of components in this app that took time to figure out the logistics for, and we haven’t reached the best solution for yet. In the future, we plan on finalizing the app for researching, testing, and implementing the AI.

One thing in the app that has not been effectively conveyed is the concept for people to avoid taking photos of faces that would promote unrealistic expectations about body images. Our current solution includes motivating quotes to remind users to turn their cameras around and for them to generate more unique words. However, this function is not apparent enough, and we hope to build a more straightforward solution in the future.