Class-agnostic object detection
AssigneeAmazon Technologies, Inc.
InventorsAyush Jaiswal, Yue Wu, Pradeep Natarajan, Premkumar Natarajan
Life of this patent
Estimating the term…
Abstract
Devices and techniques are generally described for class-agnostic object detection. In some examples, a first frame of image data comprising a first plurality of pixels may be received. First class-agnostic feature data representing the first plurality of pixels may be generated. A first object detection component may be used to determine that the first plurality of pixels corresponds to an arbitrary object represented in the first frame of image data based at least in part on the first class-agnostic feature data. Class-agnostic data indicating that the first plurality of pixels in the first frame of image data corresponds to the arbitrary object may be generated.
Claims17
A method comprising:
- receiving a first frame of image data;
- generating, by a first object detection component, first data identifying a first plurality of pixels of the first frame of image data;
- generating second data representing class-agnostic features of the first plurality of pixels;
- determining, by the first object detection component using the second data, that the first plurality of pixels corresponds to an object represented in the first frame of image data;
- receiving a natural language input referring to the object, the natural language input representing a request to perform a first action related to the object;
- receiving the second data by a natural language understanding component;
- determining, by the natural language understanding component that the natural language input corresponds to the second data;
- and
- performing the first action related to the object.
A method comprising:
- receiving a first frame of image data comprising a first plurality of pixels;
- generating first data using the first frame of image data, the first data representing at least one class-agnostic feature of a first object;
- receiving a natural language input referring to the first object;
- receiving, by a first component, the first data;
- determining, by the first component using the first data, that the first plurality of pixels corresponds to the first object represented in the first frame of image data based at least in part on the natural language input and the first data;
- and
- generating second data indicating that the first plurality of pixels in the first frame of image data includes a representation of the first object.
A system comprising:
- at least one processor;
- and
- non-transitory computer-readable memory storing instructions that, when executed by the at least one processor, are effective to:
- receive a first frame of image data comprising a first plurality of pixels;
- generate first data using the first frame of image data, the first data representing at least one class-agnostic feature of a first object;
- receive a natural language input referring to the first object;
- receiving, by a first component, the first data;
- determine, by the first component using the first data, that the first plurality of pixels corresponds to the first object represented in the first frame of image data based at least in part on the natural language input and the first data;
- and
- generate second data indicating that the first plurality of pixels in the first frame of image data includes a representation of the first object.
Dependencies are inferred from the claim text.