Region-specific image download probability modeling

Patent metadata

Field	Value
Publication number	US-9753949-B1
Application number	US-201615069246-A
Country	US
Kind code	B1
Filing date	Mar 14, 2016
Priority date	Mar 14, 2016
Publication date	Sep 5, 2017
Grant date	Sep 5, 2017

Abstract

Methods for prioritizing a set of images identified as responsive to an image search query from a user based on features of the images identified as relevant to a geographic region of the user are provided. In one aspect, the method includes submitting a plurality of images to a computer-operated convolutional neural network that is configured to analyze image pixel data for each of the plurality of images to identify features, in each of the plurality of images, influencing a download probability of the corresponding image in a plurality of geographic regions. The method also includes receiving, from the neural network and for each of the plurality of images, a download probability of each image for each of the plurality of geographic regions. Systems and machine-readable media are also provided.

First claim

What is claimed is:

1. A computer-implemented method for prioritizing a set of images identified as responsive to an image search query from a user based on features of the images identified as relevant to a geographic region of the user, the method comprising:
submitting a plurality of images to a computer-operated convolutional neural network that is configured to analyze image pixel data for each of the plurality of images to identify features, in each of the plurality of images, influencing a download probability of the corresponding image in a plurality of geographic regions; and
receiving, from the neural network and for each of the plurality of images, a download probability of each image for each of the plurality of geographic regions.

2. The method of claim 1 , further comprising:
receiving a user input from a user in one of the plurality of geographic regions comprising a search query for a collection of images;
submitting a search request comprising the search query to a search engine for the collection of images;
receiving an identification of the plurality of images from the collection of images that are responsive to the search query; and
providing a listing of the plurality of images that is prioritized according to the one of the plurality of geographic regions of the user.

3. The method of claim 2 , further comprising:
providing a first set of training images to the convolutional neural network, wherein upon processing the first set of training images the convolutional neural network is configured to identify features from an image by analyzing pixel data of the image; and
providing a second set of training images to the convolutional neural network and regional download data identifying a number of downloads of each of the second set of training images in each of the plurality of geographic regions, wherein upon processing the second set of training images and the regional download data the convolutional neural network is configured to analyze image pixel data for each of the plurality of images to identify features, in each of the plurality of images, influencing the download probability of the corresponding image in the plurality of geographic regions.

4. The method of claim 3 , wherein the providing the first set of training images and the providing the second set of training images occurs prior to the receiving the user input comprising the search query and prior to the submitting the plurality of images to the computer-operated convolutional neural network.

5. The method of claim 2 , further comprising:
providing, to the convolutional neural network with the plurality of images, information about images previously downloaded by the user,
wherein the download probability of each image comprises a download probability that is specific to the user and based on the one of the plurality of geographic regions of the user and the information about images previously downloaded by the user.

6. The method of claim 1 , further comprising:
receiving a user input from a user in one of the plurality of geographic regions comprising a search query for the plurality of images;
submitting a search request comprising the search query to a search engine for the plurality of images;
receiving an identification of a subset of the plurality of images from the collection of images that are responsive to the search query;
identifying the download probabilities for the subset of the plurality of images from the download probabilities for the plurality of images; and
providing a listing of the subset of the plurality of images that is prioritized according to the one of the plurality of geographic regions of the user.

7. The method of claim 1 , further comprising:
providing to the convolutional neural network an image that does not have associated regional download data identifying a number of downloads of the image in a geographic region; and
receiving from the convolutional neural network, a download probability of the image for each of the plurality of geographic regions.

8. A system for prioritizing a set of images identified as responsive to an image search query from a user based on features of the images identified as relevant to a geographic region of the user, the system comprising:
a memory comprising:
a first set of training images;
a second set of training images;
regional download data identifying a number of downloads of each of the second set of training images in each of the plurality of geographic regions; and
a computer-operated convolutional neural network; and

a processor configured to:
provide the first set of training images to the convolutional neural network, wherein upon processing the first set of training images the convolutional neural network is configured to identify features from an image by analyzing pixel data of the image;
provide the second set of training images to the convolutional neural network and the regional download data identifying a number of downloads of each of the second set of training images in each of the plurality of geographic regions, wherein upon processing the second set of training images and the regional download data the convolutional neural network is configured to analyze image pixel data for each of a plurality of images to identify features, in each of the plurality of images, influencing download probability of the corresponding image in the plurality of geographic regions;
submit the plurality of images to the convolutional neural network; and
receive, from the neural network and for each of the plurality of images, a download probability of each image for each of the plurality of geographic regions.

9. The system of claim 8 , wherein the processor is further configured to:
receive a user input from a user in one of the plurality of geographic regions comprising a search query for a collection of images;
submit a search request comprising the search query to a search engine for the collection of images;
receive an identification of the plurality of images from the collection of images that are responsive to the search query; and
provide a listing of the plurality of images that is prioritized according to the one of the plurality of geographic regions of the user.

10. The system of claim 9 , wherein the providing the first set of training images and the providing the second set of training images occurs prior to the receiving the user input comprising the search query and prior to the submitting the plurality of images to the computer-operated convolutional neural network.

11. The system of claim 9 , wherein the processor is further configured to:
provide, to the convolutional neural network with the plurality of images, information about images previously downloaded by the user,
wherein the download probability of each image comprises a download probability that is specific to the user and based on the one of the plurality of geographic regions of the user and the information about images previously downloaded by the user.

12. The system of claim 8 , wherein the processor is further configured to:
receive a user input from a user in one of the plurality of geographic regions comprising a search query for the plurality of images;
submit a search request comprising the search query to a search engine for the plurality of images;
receive an identification of a subset of the plurality of images from the collection of images that are responsive to the search query;
identify the download probabilities for the subset of the plurality of images from the download probabilities for the plurality of images; and
provide a listing of the subset of the plurality of images that

Assignees

• Shutterstock Inc

Inventors

• Malpani Vaibhav
• Hurst Nathan

Classifications

• G06F16/252

  between a Database Management System and a front-end application · CPC title

• G06F16/583

  using metadata automatically derived from the content · CPC title

• G06N3/084

  Backpropagation, e.g. using gradient descent · CPC title

• G06F16/24578

  using ranking · CPC title

• G06F16/587 Primary

  using geographical or spatial information, e.g. location · CPC title