US 2019/0195634 A1 · CUSTOMER SERVICE SYSTEMS AND PORTALS

Abstract

A device, system, and methods are described to perform machine-learning camera-based indoor mobile positioning. The indoor mobile positioning may utilize inexact computing, wherein a small decrease in accuracy is used to obtain significant computational efficiency. Hence, the positioning may be performed using a smaller memory overhead at a faster rate and with lower energy cost than previous implementations. The positioning may not involve any communication (or data transfer) with any other device or the cloud, providing privacy and security to the device. A hashing-based image matching algorithm may be used which is cheaper, both in energy and computation cost, over existing state-of-the-art matching techniques. This significant reduction allows end-to-end computation to be performed locally on the mobile device. The ability to run the complete algorithm on the mobile device may eliminate the need for the cloud, resulting in a privacy-preserving localization algorithm by design since network communication with other devices may not be required.

Claims20

Dependent claims

A method for machine-learning camera-based indoor mobile positioning, the method comprising:

by a user equipment device (UE):
for each of a plurality of respective training images, receiving a first hashing-based mapping from a first plurality of multidimensional features of each respective training image to entries in a key index;
receiving a query image;
extracting a second plurality of multidimensional features from the query image;
performing a second hashing-based mapping from each of the second plurality of multidimensional features to entries in the key index;
identifying a first training image of the plurality of training images, wherein the first training image is identified based on a correspondence between the respective first hashing-based mapping of the first training image and the second hashing-based mapping;
and
determining a location of the UE based on a location tag associated with the first training image.

5 more elements

A user equipment device (UE), comprising:

a camera;
a memory medium;
and
digital circuitry coupled to the camera and the memory medium;
wherein the camera is configured to acquire a query image;
wherein the digital circuitry is configured to:
for each of a plurality of respective training images, receive a first hashing-based mapping from a first plurality of multidimensional features of each respective training image to entries of a key index, and store the first hashing-based mappings in the memory medium;
receive the query image from the camera;
extract a second plurality of multidimensional features from the query image;
perform a hashing-based mapping from each of the second plurality of multidimensional features to entries of the key index;
for each of the respective training images, determine a number of multidimensional features of the first plurality of multidimensional features of the respective training image that are mapped to the same entries of the key index as any of the second plurality of multidimensional features;
identify a first training image of the plurality of training images that has the largest number of multidimensional features that are mapped to the same entries of the key index as any of the second plurality of multidimensional features;
determine a location of the UE based on a location tag associated with the first training image.

10 more elements

A non-transitory computer-readable memory medium configured to store program instructions which, when executed, are configured to cause a user equipment device (UE) to:

for each of a plurality of respective training images, receive a first hashing-based mapping from a first plurality of multidimensional features of each respective training image to entries in a key index, and store the first hashing-based mapping in the memory medium;
receive a query image;
extract a second plurality of multidimensional features from the query image;
perform a second hashing-based mapping from each of the second plurality of multidimensional features to entries in the key index;
identify a first training image of the plurality of training images, wherein the first training image is identified based on a correspondence between the respective first hashing-based mapping of the first training image and the second hashing-based mapping;
and
determine a location of the UE based on a location tag associated with the first training image.

4 more elements

Dependencies are inferred from the claim text.

Customer service systems and portals

Life of this patent

Customer service systems and portals

Life of this patent

Abstract

Claims20