Systems, apparatus, and methods to improve object recognition
AssigneeIntel Corporation
InventorOleg Pogorelik
Life of this patent
Estimating the term…
Abstract
Methods, apparatus, systems, and articles of manufacture are disclosed to improve robot object recognition. An example apparatus includes a visual object recognizer to obtain a visual identifier associated with a target object, and a recognizable object model generator to generate a model of the target object based on mapping an image of the target object to classifier information corresponding to the visual identifier.
Claims20
An apparatus to perform object-recognition, the apparatus comprising:
- a visual object recognizer to compare a first image of a first target object to one or more first models;
- a visual identifier recognizer to utilize a uniform resource locator (URL) to obtain classifier information associated with the first target object, the URL based on a visual identifier on the first target object;
- and
- a recognizable object model generator to generate a second model of the first target object based on the classifier information;
- the visual object recognizer to:
- obtain a second image of a second target object, the second target object to be substantially similar to the first target object, the second target object to be without the visual identifier;
- and
- identify the second target object by comparing the second image to the second model.
A non-transitory computer readable storage medium comprising instructions which, when executed, cause a machine to at least:
- compare a first image of a first target object to one or more first models;
- obtain classifier information associated with the first target object via a uniform resource locator (URL), the URL based on a visual identifier on the first target object;
- generate a second model of the first target object based on the classifier information;
- obtain a second image of a second target object, the second target object substantially similar to the first target object, the second target object to be without the visual identifier;
- and
- identify the second target object by processing the second image using the second model.
A method to perform object-recognition, the method comprising:
- comparing, by executing an instruction with a processor of a device, a first image of a first target object to one or more first models;
- obtaining, by executing an instruction with the processor of the device, classifier information associated with the first target object via a uniform resource locator (URL), the URL based on a visual identifier on the first target object;
- and
- generating, by executing an instruction with the processor of the device, a second model of the first target object based on the classifier information;
- obtaining a second image of a second target object, the second target object substantially similar to the first target object, the second target object to be without the visual identifier;
- and
- identifying the second target object based on a comparison of the second image to the second model.
An apparatus to perform object-recognition, the apparatus comprising:
- a visual object recognizer to:
- compare a first image of a first target object to one or more first models;
- direct a manipulation of a second target object to capture a portion of the second target object in a second image, the second target object to be substantially similar to the first target object, the second target object to be without a visual identifier, the portion to include the visual identifier;
- and
- identify the second target object by comparing the second image to a second model of the first target object;
- a visual identifier recognizer to obtain classifier information associated with the first target object based on a uniform resource locator (URL), the URL based on the visual identifier on the first target object;
- and
- a recognizable object model generator to generate the second model based on the classifier information.
Dependencies are inferred from the claim text.