Object re-identification

A non-transitory computer-readable medium storing one or more instructions that, when executed by one or more processors, are configured to cause the one or more processors to perform operations comprising:

• obtaining a first image from a first sensor;
• detecting a first object within the first image;
• extracting discriminative features of the first object into a first feature vector;
• comparing the first feature vector to each vector cluster of a plurality of vector clusters composing a feature vector database, each vector cluster of the plurality of vector clusters having a unique identifier;
• associating the first feature vector with a first unique identifier of a first vector cluster when a comparison threshold is met as a result of the comparing the first feature vector to the first vector cluster of the plurality of vector clusters;
• obtaining a second image from a second sensor;
• detecting a second object within the second image;
• extracting discriminative features of the second object into a second feature vector;
• comparing the second feature vector to each vector cluster of the plurality of vector clusters composing the feature vector database;
• associating the second feature vector with the first unique identifier when the comparison threshold is met as a result of the comparing the second feature vector to the first vector cluster of the plurality of vector clusters;
• and
• determining, based on the association of the second feature vector with the first unique identifier, that the second object is same as the first object by virtue of the relationship of the first feature vector and the second feature vector to the first vector cluster.

A system comprising:

• one or more processors;
• and
• one or more computer-readable media storing instructions that, when executed by one or more processors, are configured to cause the one or more processors to perform operations comprising:
• obtaining a first image from a first sensor;
• detecting a first object within the first image;
• extracting discriminative features of the first object into a first feature vector;
• comparing the first feature vector to each vector cluster of a plurality of vector clusters composing a feature vector database, each vector cluster of the plurality of vector clusters having a unique identifier;
• associating the first feature vector with a first unique identifier of a first vector cluster when a comparison threshold is met as a result of the comparing the first feature vector to the first vector cluster of the plurality of vector clusters;
• obtaining a second image from a second sensor;
• detecting a second object within the second image;
• extracting discriminative features of the second object into a second feature vector;
• comparing the second feature vector to each vector cluster of the plurality of vector clusters composing the feature vector database;
• associating the second feature vector with the first unique identifier when the comparison threshold is met as a result of the comparing the second feature vector to the first vector cluster of the plurality of vector clusters;
• and
• determining, based on the association of the second feature vector with the first unique identifier, that the second object is same as the first object by virtue of the relationship of the first feature vector and the second feature vector to the first vector cluster.