US 11,113,841 B1Granted

Self-learning three-dimensional boundaries from images

AssigneeAmazon Technologies, Inc.

InventorsLev Zelenskiy, Jon Ducrou, James Eather, Michael Neville

Life of this patent

Estimating the term…

Loading patent…

Abstract

Techniques are disclosed. The techniques include receiving, from a camera, an image showing a portion of a physical topography and target object. The techniques include determining, based on the image, a location of the target object shown in the image and a physical characteristic of the target object shown in the image, the physical characteristic being common with at least another target object. The techniques then include determining a three dimensional (3D) position of the target object in a 3D representation of the portion of the physical topography, the 3D position determined based on (i) the location of the target object, and (ii) the physical characteristic. Based on detecting a cluster of 3D positions of target objects (over several images), the techniques involve generating a 3D boundary of a zone of a portion of the physical topography.

Claims20

Dependent claims

A system, comprising:

a camera;
and
a computer system communicatively coupled with the camera, the computer system comprising one or more processors and one or more computer-readable storage media storing instructions that, upon execution by the one or more processors, cause the computer system to, at least:
receive, from the camera, a first image of a first plurality of images representing a motion of a first vehicle license plate through a portion of a parking lot, the portion being within a field of view of the camera;
determine, based at least in part on the first image, a location of the first vehicle license plate shown in the first image;
determine, based at least in part on the first image, a physical characteristic of the first vehicle license plate shown in the first image, the physical characteristic being common with a second vehicle license plate;
generate a three-dimensional (3D) representation of the portion of the parking lot based at least in part on (i) the location and (ii) the physical characteristic;
subsequent to generating the 3D representation of the portion of the parking lot, receive, from the camera, a second image of a second plurality of images, the second image showing the portion of the parking lot and the second vehicle license plate;
determine a 3D position of the second vehicle license plate in the 3D representation based at least in part on (i) a second location of the second vehicle license plate shown in the second image, and (ii) the physical characteristic shown in the second image;
determine that a vehicle is parked in the parking lot, the vehicle being associated with the second vehicle license plate;
based in part on determining that the vehicle is parked, determine that the 3D position belongs to a cluster of 3D positions of one or more vehicle license plates shown in the second plurality of images;
and
generate a 3D boundary of a parking space within the portion of the parking lot based at least in part on the 3D positions of the cluster, the parking space corresponding to where the vehicle is parked.

10 more elements

A method implemented on a computer system, the method comprising:

receiving, from a camera, an image showing a portion of a parking lot and a vehicle license plate;
determining, based at least in part on the image, a location of the vehicle license plate shown in the image;
determining, based at least in part on the image, a physical characteristic of the vehicle license plate shown in the image, the physical characteristic being common with at least another vehicle license plate;
determining a three dimensional (3D) position of the vehicle license plate in a 3D representation of the portion of the parking lot, the 3D position determined based at least in part on (i) the location of the vehicle license plate, and (ii) the physical characteristic;
determining that the 3D position belongs to a cluster of 3D positions of one or more vehicle license plates;
and
generating a 3D boundary of a parking space for a vehicle within the portion of the parking lot based at least in part on the 3D positions of the cluster.

4 more elements

A non-transitory computer-readable storage medium storing instructions that, upon execution on a computer system, configure the computer system to perform operations comprising:

receiving, from a camera, an image showing a portion of a physical topography and a target object, the physical topography being divided into a plurality of zones;
determining, based at least in part on the image, a location of the target object shown in the image;
determining, based at least in part on the image, a physical characteristic of the target object shown in the image, the physical characteristic being common with at least another target object;
determining a three dimensional (3D) position of the target object in a 3D representation of the portion of the physical topography, the 3D position determined based at least in part on (i) the location of the target object, and (iii) the physical characteristic;
determining that the 3D position belongs to a cluster of 3D positions of one or more target objects;
and
generating a 3D boundary of a zone of the plurality of zones of the portion of the physical topography based at least in part on the 3D positions of the cluster, the zone corresponding to the 3D position of the target object.

4 more elements

Dependencies are inferred from the claim text.