How full is it right now?

Study Spot counts people at the door, so you see open seats.

Demo built from real counts on a public sample clip.
In
0
Out
0
Inside now
284
Open seats
16 of 300
0:00 / 0:00

Where it can help

Places that get too crowded. Count every door or gate, and staff see how full each part is before it is unsafe.

A two storey reading room with rows of desks

Libraries in exam week

See which floor has open seats before you walk over.

The app is built for this.

  • Stadiums and arenas

    See which gates have short lines on game day, and send fans to the fastest way in.

    A full stadium crowd at a night football game
  • Concerts and festivals

    Know how many people are in each area, from the main stage to the food court.

    A big crowd in front of a lit stage at night
  • Stations and platforms

    See a platform filling up early, while there is time to act. I have run it on a busy subway clip.

    Crowded subway platforms on both sides of the track
  • Gyms and rec centers

    See how busy it is right now.

    A gym with a row of cardio machines by the windows

One camera per door. Every space, at a glance.

In an emergency, it shows how many may still be inside.

Responders see how many went in and how many came out.

Responders get a clear number to check against their roll call.

Went in
0
Came out
0
May still be inside
284

Demo counts, from the doorway at the top of the page.

How it counts

  1. A camera looks down at a doorway.
  2. YOLO, an AI model, finds each body, never faces.
  3. ByteTrack follows each one for a few seconds.
  4. Crossing the line is one in or one out.

Inside now in minus out

Open seats seats minus inside

Counts reset at 2 am each night. A building adds up its floors.

What it keeps

People in. People out. The time.

What it never keeps

Faces
It finds body shapes, not who someone is.
Video
Frames stay in memory and are never saved.
Names
It can say how many, never who.

YOLO got 60 of 63 crossings right.

The older OpenCV HOG detector got 8. To check them, I counted every crossing in the test clips myself.

Crossings in and out, by hand and by each counter
ClipBy hand, in and outYOLOOpenCV HOG
Hallway, lower line9 / 109 / 103 / 3
Hallway, upper line10 / 1310 / 131 / 0
Market square, from above9 / 1111 / 100 / 0
Grocery store0 / 10 / 10 / 1

YOLO26 nano at 1280 px with ByteTrack, about 37 to 40 frames a second on a laptop GPU (Apple M1).

The app’s campus list with each library and its open seats, on simulated data
Hayden Library in the app, seats open by floor, on simulated data

What is real today

Real
The counter works on recorded video.
Next
Cameras on real doors. Until then, the app runs on simulated data.

Campus list, floors and zones, Vacant or Full, and Find me a seat for quiet or group study.

Open the app

I pitched it at ASU.

Presenting the problem slide: finding a seat at Hayden or Noble can feel impossible
The problem: every floor is packed.
Presenting the solution slide: the Study Spot app showing open seats by library and by floor
The solution: open seats, floor by floor.