For judging

If the iyifx.kndrck.co isn't working, please clone the repository and open the docs/ folder and open index.html

If you wanna test the live site, use a public instagram username, or one of the following (for testing):

  • mr.miso.oz (good boy cat with no depression)
  • harryquin56 (some dude with questionable posts)

Inspiration

According to Time.com, instagram is the worse social media platform for mental health. And it's easy to see why: you’re constantly being bombarded with how perfect everyone else’s life is, and it's easy to overthink and self doubt, leading to bad mental health down the road

What it does

IYIFX helps identify your friends’ mental state purely based on instagram posts, so you can provide support in their time of need.

How We built it

To accomplish what we did today, we had to solve 3 problems:

  1. We wanted to obtain instagram images with an API key, and to do that we reverse engineered Instagram’s API
  2. We were facing CORS error while trying to do a PWA with our reversed engineered API, as such we used GCP’s cloud function to scrap and pass back our data
  3. We needed a way to determine if a photo showed mental health symptoms, the clarifai API was used to extract meaningful data from images, the hand picked features (based on a research paper) were then passed down a decision tree.

Challenges We ran into

We had to manually hand tune the extracted features from Clarify in order to better generalize classification

We ran into CORs issues when we used our reversed engineered API when we used the PWA

Accomplishments that I'm proud of

Managed to reverse engineer Instagram's API and overcome CORs issues with the reversed engineered API via google cloud functions

What I learned

How to use clarifai and Google Cloud functions

What's next for IYIFX

Using NLP to extract sentiment from a user's Instagram caption

Built With

Share this project:
×

Updates