PeakScroll

Three kind fellas peak scrolling.

Inspiration

Following from the theme of mash-up mania, we wanted to tie together two disconnected ideas: the unhealthiness in doom-scrolling, and healthy posture and more expressive interaction. PeakScroll is a result of this combination, as it allows the act of doom scrolling to be more engaging and interactive, and hopefully healthier; no more doom-scrolling, start peak-scrolling!

What it does

PeakScroll uses real-time emotion detection through your webcam to automatically scroll Instagram Reels or YouTube Shorts based on your mood and gestures.

How we built it

We combined Python, OpenCV, and a Flask server with Chrome DevTools Protocol automation, plus a lightweight HTML/CSS dashboard for feedback.

Challenges we ran into

Dependency hell when juggling both gesture CV models with emotion-analyzing CV models.

Accomplishments that we're proud of

We were able to create an MVP within just a couple hours, and polished our system without rushing. It's also very enjoyable to use, which was surprising to us.

What we learned

We learned how powerful it is to interact with computers using gestures and tangible interfaces, as opposed to pure physical interaction. Additionally, CV technology has progressed to where we can successfully run this project on a relatively mid-range laptop, with no cloud processing.