Recent FB- Cambridge Analytica scandal. We thought wouldn't it be great if we could test peoples Instagram account and look for specific traits. Such as if an accounts is fake or if the account has potential to become a paid influecer or model.

What it does enables us to get the access token for users Instagram accounts. With the access token we collect users images, media and statistics. With this we run some basic ratio models to learn of how engaged the users audience is. Then we take all of the users images and run it trough our Visual Recognition model. With this the aim is to detect potential influencers/models.

How we built it

Most of it is built in Python, with some minor data analysis parts in R.

Challenges we ran into

Getting the Instagram users access token. Since we are hosting our domain for free, we only had

Accomplishments that we're proud of

Getting the access token. Training our Visual Recognition model with over 1000 images.

What we learned

First time exposure to IBM Watson, thanks to Mofi. First time Hackathon, first time not sleeping for 24 hours which we believed would hurt us but we seem to be doing fine.

What's next for Social Media Analysis

Move to a real hosting that can provide us instant logs of users codes.

Share this project: