Inspiration
We wanted to make Instagram scrolling more engaging by turning it into a game. Everyone loves making predictions, and Instagram reels provide the perfect platform for a quick, fun guessing game.
What it does
Reel Analysis turns Instagram scrolling into an interactive estimating competition. Users predict how many likes a reel will have in an hour, then check back (or keep scrolling for an hour) to see how close they were.
As part of our vision to create a more engaging and customized social media experience, we designed a second component for x.com that uses machine learning to detect and replace any image of a person with custom images of your choice. This has applications for removing people like ex's or celebrities that you no longer want to see. For our example, we remove images of people who can't sing (bababouie) with images of someone who can (Youngboy Never Broke Again).
How we built it
Created a lightweight Chrome extension with a Flask backend. The extension adds a prediction interface directly to Instagram reels, while the backend tracks predictions and calculates accuracy scores.
Challenges we ran into
Making the UI feel natural within Instagram's interface (e.g. we made sure comments were still visible) and ensuring the game-like experience doesn't interfere with normal browsing. Also had to handle Instagram's dynamic content loading smoothly. Fetching data using instagrapi was also a challenge due to rate-limiting issues by Meta (involving over 3 banned Instagram accounts).
What we learned
How to create engaging, game-like experiences within existing platforms. Also learned about balancing functionality with user experience in browser extensions.
What's next for Reel Analysis
Add a leaderboard system, achievement badges for accurate predictions, and social features to compete with friends. Maybe even add different game modes or prediction categories. We also are interested in integrating components into a single extension available across social media platforms.
Built With
- flask
- instagrapi
- javascript
- python
Log in or sign up for Devpost to join the conversation.