Friday, November 6, 2026

From 2D videos to a live 3D map

Every camera around your home and everything it sees, together in one live view.

We’re inventing new ways to help people make sense of what’s happening around their home.

Multiple cameras together can capture all the action: a video doorbell at the front door, a camera over the driveway, another facing the backyard. But every camera records its own 2D videos, and it’s really hard for humans to piece them together. To find out where a delivery driver went, you jump between clips from different cameras and have to reconstruct the route in your head.

A live 3D map that shows how the delivery driver walks around your porch is much more intuitive. So that’s what we’ve built at Plumerai.

Just enter your address

Our AI starts by automatically positioning your cameras on a 3D map of your home. All you need to do is enter your address. The map of your home is built and your cameras are located on it, with no need to walk around the house and scan it with your phone.

One view instead of many clips

Once the map is ready, the people and vehicles your cameras see show up on it: a delivery driver, the van parked at the curb, two people walking a dog down the sidewalk. You can follow a person as they move from one camera’s view to the next and see their whole route around the house at a glance, instead of stitching together clips from several cameras in your mind. It’s a more intuitive and much faster way of understanding what’s happening around your home.

The 3D model isn’t a fixed picture, either. You can rotate it, zoom in, and look at your home and the activity around it from any angle.

Following people from camera to camera

To show one continuous route, the map has to know that the person walking up the driveway is the same person who appears at the front door a few seconds later, on a different camera and from a different angle. That’s what Plumerai Multi-Camera Re-Identification does. It recognizes the same person across cameras based on their overall appearance, such as clothing and build, so it works even when their face isn’t visible. We’ve written before about what it took to make re-identification accurate, and on the 3D map it turns detections from separate cameras into the path of a single person.

What happened, in words and in video

We combine the map with Plumerai AI Captions, which describe what happened in plain language. Every event comes with a short description, such as “An Amazon delivery person delivered packages to the house,” so you can tell what happened at a glance.

The original video doesn’t disappear, though. The 3D scene still shows portions of the video clips, cropped around the person or vehicle and placed right next to them on the map. You get the overview of the 3D map and the detail of the real footage in a single view.

Standard cameras, no special sensors

It all works with standard smart home cameras. There’s no radar or any other special sensor involved. Everything is done by the AI running on the standard video recordings. That means it doesn’t require any hardware changes and can be rolled out with just a software update.

Contact

Building smart home or security cameras? Contact us to learn how a live 3D map can work with your cameras.