In Depth

Here’s everything IsMomOnTheFloor does, in depth.

The camera checking schedule

This clockface shows when we look through each camera.

0 5 10 15 20 25 30 35 40 45 50 55 one hour
AI Level 1 AI Level 2 extra looks during motion

Each camera is checked every 6 minutes, plus extra checks when motion is detected. In total, we look through each camera over 7,000 times a month. The combination of clockwork looks plus motion-activated looks means any problem is more likely to be seen.

Every time we connect to a camera, our system requests a “Live View”. We don’t want video but it’s the only way to connect. We take the first frame, analyze it, and delete it. In your Ring app, you will see Live View clips every 6 minutes. We never see them.

Ring app screenshot: the camera’s history timeline with a Live View at 11:00:54 AM Ring app screenshot: the same camera’s next Live View, at 11:06:48 AM Ring app screenshot: the same camera’s next Live View, at 11:12:56 AM

Two layers of AI

Different AI systems have different blind spots. Two independent sets of eyes see more than one.

AI Level 1

The Generalist

Finds most examples of someone on the floor.

AI Level 2

The Specialist

Finds some edge cases that Level 1 misses.

System Accuracy & Alert Strictness

How well does our system see someone on the floor? And should we always send an alert?

Alert Strictness

Important: You want to know if someone is on the floor, but you don’t want too many false alarms. This controls that balance.

1 - Slightly strict
  1. Send an alert as soon as one AI thinks someone is on the floor. May generate some false alarms, especially transient ones.
  2. Look again after ~20 seconds before sending an alert. Eliminates transient false alarms.
  3. Look again after ~20 seconds and ~40 seconds before sending an alert. Eliminates longer transient false alarms.
  4. Two AIs must agree that someone is on the floor, and look again after ~20 seconds, before sending an alert.
  5. Two AIs must agree that someone is on the floor, and look again after ~20 seconds and ~40 seconds, before sending an alert.

As you increase the Strictness level, alerts will take a little longer or may not happen at all.

The AIs are amazing at analyzing images but they can make mistakes. Sometimes they see a person briefly on the floor to pick up something, which is real but probably not worth an alert. And sometimes when using night vision, the AIs might mistake a pile of pillows on the floor for a person.

So we developed Alert Strictness: We ask the AIs to look multiple times, or to compare with each other, before sending an alert.

Sometimes the best course of action is to move a camera for a better angle, or to reduce clutter in the monitored area.

System Accuracy

As of October, 2026, we have a growing “corpus” of 204 images that span 3 test homes. These images show us on the floor in many different places and types of lighting. The corpus also includes images where we know that no one is on the floor… either the room is empty, or there’s a large dog, etc.

Here’s how our system performs at spotting us on the floor (called “recall”), as well as how often it thinks someone is on the floor when no one actually is (“false positive”):

  • At Alert Strictness level 0: Recall is 96.8%; false positives are 3.8%
  • At Alert Strictness level 4: Recall is 94.4%; false positives are 0%

Your results may be different. These figures are from our 3 test homes, with our cameras, rooms, and lighting.

Test it yourself, any time

Lie down and wait 1-6 minutes for an alert. Or start a test anytime on your account page:

Test the system

Get into position on the floor first, then press the button. We take an immediate look with the same checks as real monitoring, and you’ll get a TEST RESULT email either way — detected, not detected, or why we couldn’t check.

Test now — I’m on the floor

The actual control on your account page.

The test runs the real thing: a fresh image from each camera, judged by both AI levels, through the same alert path that real monitoring uses.

The TEST RESULT email shows each AI level’s verdict and the image we captured — so if a camera’s angle can’t see you on the floor behind the couch, you’ll know as quickly as possible.

This is useful any time you want to see what the camera sees and what the AIs perceive.

Emails that keep you posted

In addition to alerts, here are some other emails you’ll receive:

Once-a-day status email — positive proof that the system is up, so silence is never ambiguous.

A camera that goes quiet never goes quietly: you hear about it, once, and you hear when it’s back — no email bombardment in between.

Your account page keeps a record

A running record of recent checks — outcomes and times only, no images:

Recent activity

Last check: 4 minutes ago — no one on the floor

  • 2 hours ago: Camera “Living Room”: Test check: ALERT sent (possible person on the floor) AI Level 1: Someone might be on the floor AI Level 2: Someone might be on the floor
  • Yesterday: Camera “Kitchen”: Motion check: possible person on the floor — covered by the recent alert

The activity record on your account page.

Every check that saw something is recorded — when it ran, what triggered it (scheduled, motion, or a test), and what each AI level concluded. Routine all-clear checks just update the “last check” line.

Also included

Second person — add a 2nd person to receive alerts if mom is on the floor.
Night coverage — checks continue in the dark using the cameras’ infrared night vision.
No new hardware — works with the Ring cameras you already own.
No humans look at images — we get a camera frame and pass it to our AI provider for instant analysis by machines.
Consent built in — recording the monitored person’s consent is part of setup, not an afterthought.
Billed through Ring — monthly, per camera, cancel any time in the Ring app.
Get started See About Us →