Information processing apparatus and information processing method

• US 10,255,517 B2
• Filed: 03/24/2015
• Issued: 04/09/2019
• Est. Priority Date: 03/27/2014
• Status: Active Grant

• Current Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Original Assignee: Canon Kabushiki Kaisha (Canon Inc.)
• Inventors: Uchiyama, Hiroyuki
• Primary Examiner(s): Couso, Yon J
• Application Number:
  US15/121,899
• Publication Number:
  US 20170017859A1
• Time in Patent Office: 1,477 Days
• CPC Class Codes: G06F 18/214 Generating training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06F 18/2148 characterised by the process organisation or structure, e.g. boosting cascade
• CPC Class Codes: G06F 18/254 of classification results, e.g. of results related to same input data
• CPC Class Codes: G06F 18/256 of results relating to different input data, e.g. multimodal recognition
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: G06V 10/774 Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06V 10/7747 Organisation of the process, e.g. bagging or boosting
• CPC Class Codes: G06V 10/809 of classification results, e.g. where the classifiers operate on the same input data
• CPC Class Codes: G06V 10/811 the classifiers operating on different input data, e.g. multi-modal recognition
• CPC Class Codes: G06V 20/52 Surveillance or monitoring of activities, e.g. for recognising suspicious objects recognising microscopic objects G06V20/69
• CPC Class Codes: G06V 40/103 Static body considered as a whole, e.g. static pedestrian or occupant recognition

Abstract

A feature amount, and context information representing a context regarding an image are obtained from the image. The first likelihood indicating the likelihood at which the image is an image of a specific object is obtained from the feature amount. The second likelihood indicating the likelihood at which the image is an image of the specific object is obtained from the context information. Whether the image is an image of the specific object is identified using the first likelihood and the second likelihood.

Claims

1. An information processing apparatus comprising:

one or more processors; and

a memory coupled to the one or more processors, the memory having stored thereon instructions which, when executed by the one or more processors, cause the apparatus to;

obtain a feature amount from an input image;

obtain context information representing a context regarding the input image, wherein the context information includes a position and size of the input image in an original image from which the input image has been extracted;

obtain, from the feature amount, a first likelihood indicating a likelihood at which the input image is an image of a specific object;

obtain, from the context information, a second likelihood indicating a likelihood at which the input image is an image of the specific object; and

identify, by using the first likelihood and the second likelihood, whether the input image is an image of the specific object,

wherein the obtaining of the second likelihood is performed by a discriminator which has been learned using the first likelihood and the context information.

2. The apparatus according to claim 1, wherein the discriminator has been selected from candidates of the discriminator generated using the first likelihood and the context information.

3. The apparatus according to claim 2, wherein the candidates of the discriminator have been generated further using a weight set based on the first likelihood for a sample obtained by joining the feature amount and the context information.

4. An information processing apparatus comprising:

one or more processors; and

a memory coupled to the one or more processors, the memory having stored thereon instructions which, when executed by the one or more processors, cause the apparatus to;

obtain a feature amount from an input image;

obtain context information representing a context regarding the input image, wherein the context information includes a position and size of the input image in an original image from which the input image has been extracted;

obtain, from the feature amount, a first likelihood indicating a likelihood at which the input image is an image of a specific object; and

learn a discriminator being different from a discriminator for obtaining the first likelihood from the feature amount by using the first likelihood and the context information.

5. The apparatus according to claim 4, wherein in the learning, the instructions, when executed by the one or more processors, cause the information processing apparatus to:

generate candidates of a discriminator by using the first likelihood and the context information; and

select at least one candidate from the candidates of the discriminator based on an evaluation score obtained for each of the candidates of the discriminator.

6. An information processing method for an information processing apparatus, the information processing method comprising:

obtaining a feature amount from an input image;

obtaining context information of the information processing apparatus, to obtain context information representing a context regarding the input image, wherein the context information includes a position and size of the input image in an original image from which the input image has been extracted;

obtaining, from the feature amount, a first likelihood indicating a likelihood at which the input image is an image of a specific object;

obtaining, from the context information, a second likelihood indicating a likelihood at which the input image is an image of the specific object; and

identifying, by using the first likelihood and the second likelihood, whether the image is an image of the specific object,

wherein the obtaining of the second likelihood is performed by a discriminator which has been learned using the first likelihood and the context information.

7. An information processing method for an information processing apparatus, the information processing method, comprising:

obtaining a feature amount from an input image;

obtaining context information representing a context regarding the input image, wherein the context information includes a position and size of the input image in an original image form which the input image has been extracted;

obtaining, from the feature amount, a first likelihood indicating a likelihood at which the input image is an image of a specific object; and

learning a discriminator being different from a discriminator for obtaining the first likelihood from the feature amount by using the first likelihood and the context information.

8. A non-transitory computer-readable storage medium storing a computer program for causing a computer to execute a method of controlling an information processing apparatus, the method comprising:

obtaining a feature amount from an input image;

obtaining context information representing a context regarding the input image, wherein the context information includes a position and size of the input image in an original image from which the input image has been extracted;

obtaining, from the feature amount, a first likelihood indicating a likelihood at which the input image is an image of a specific object;

obtaining, from the context information, a second likelihood indicating a likelihood at which the input image is an image of the specific object; and

identifying, by using the first likelihood and the second likelihood, whether the input image is an image of the specific object,

wherein the obtaining of the second likelihood is performed by a discriminator which has been learned using the first likelihood and the context information.

9. A non-transitory computer-readable storage medium storing a computer program for causing a computer to execute a method of controlling an information processing apparatus, the method comprising:

obtaining a feature amount from an input image;

obtaining context information representing a context regarding the input image, wherein the context information includes a position and size of the input image in an original image form which the input image has been extracted;

obtaining, from the feature amount, a first likelihood indicating a likelihood at which the input image is an image of a specific object; and

learning a discriminator being different from a discriminator for obtaining the first likelihood from the feature amount by using the first likelihood and the context information.