Visual object and event detection and prediction system using saccades
AssigneeInternational Business Machines Corporation
InventorsBAN KAWAS, ARVIND KUMAR, JANUSZ MARECKI, SHARATHCHANDRA U. PANKANTI
Life of this patent
Estimating the term…
Abstract
A method of operating an image detection device includes receiving an image, dividing the image into a plurality of patches, grouping ones of the plurality of patches, generating a set of saccadic paths through the plurality of patches of the image, generating a cluster-direction sequence for each saccadic path, generating a policy function for identifying an object in a new image using a combination of the cluster-direction sequences, and operating the image detection device using the policy function to identify an object in the new image.
Claims10
An image detection device comprising:
- a memory;
- and
- at least one processor, coupled to said memory, and operative to:
- receive an image;
- divide the image into a plurality of patches;
- group ones of the plurality of patches;
- generate a set of saccadic paths through the plurality of patches of the image;
- generate a cluster-direction sequence for each saccadic path;
- generate a policy function for identifying an object in a new image using a combination of the cluster-direction sequences;
- and
- operate the image detection device using the policy function to identify an object in the new image.
An image detection device comprising:
- a memory;
- and
- at least one processor, coupled to said memory, and operative to:
- receive an image;
- divide the image into a plurality of patches;
- generate a cluster-direction sequence for each of a plurality of saccadic paths to be explored, wherein the plurality of saccadic paths are given by a policy matrix;
- and
- identify an object in the image using the cluster-direction sequences.
The image detection device of claim 9, wherein the at least one processor is operative to weight by using a total frequency of the occurrence of the cluster-direction sequence in the policy matrix for a given class defined by the policy matrix.
Dependencies are inferred from the claim text.