Method for unsupervised clustering of multimedia data using a large-scale matching system
AssigneeCortica, Ltd.
InventorsIgal Raichelgauz, Karina Odinaev, Yehoshua Y. Zeevi
Life of this patent
Estimating the term…
Abstract
A method and system for unsupervised clustering of multimedia content are provided. The method includes generating a plurality of clusters, each cluster containing at least a data element; generating for each of the plurality of clusters a corresponding signature; matching each of the signatures to all other signatures; determining a clustering score for each match; clustering multimedia data elements of each pair of clusters of the plurality of clusters that are determined to have a clustering score above a threshold value to create at least a first cluster; and storing the at least a first cluster in a storage unit.
Claims22
A computerized method for unsupervised clustering, comprising:
- generating a plurality of clusters, each cluster containing at least a data element;
- generating for each of the plurality of clusters a corresponding signature, wherein a signature is generated from multiple patches of a multimedia data element, wherein multiple patches are of random length and random position within the multimedia data element;
- matching each of the signatures to all other signatures;
- determining a clustering score for each match;
- clustering multimedia data elements of each pair of clusters of the plurality of clusters that are determined to have a clustering score above a threshold value to create at least a first cluster;
- and
- storing the at least a first cluster in a storage unit.
An apparatus for unsupervised clustering, comprising:
- at least one processing unit;
- a storage unit for storing one or more clusters of multimedia data elements;
- a memory coupled to the at least one processing unit, the memory contains instructions that, when executed by the at least one processing unit configure the apparatus to:
- generate a plurality of clusters, each cluster containing at least a data element;
- generate for each of the plurality of clusters a corresponding signature, wherein a signature is generated from multiple patches of a multimedia data element, wherein multiple patches are of random length and random position within the multimedia data element;
- match each of the signatures to all other signatures;
- determine a clustering score for each match being performed;
- cluster multimedia data elements of each pair of clusters of the plurality of clusters having a clustering score above a threshold value to create at least a first cluster;
- and
- store the at least a first cluster in the storage unit.
Dependencies are inferred from the claim text.