Scene recognition method and apparatus

Patent metadata

Field	Value
Publication number	US-9465992-B2
Application number	US-201514657121-A
Country	US
Kind code	B2
Filing date	Mar 13, 2015
Priority date	Sep 14, 2012
Publication date	Oct 11, 2016
Grant date	Oct 11, 2016

Abstract

A scene recognition method and apparatus are provided. The method includes obtaining multiple local detectors by training a training image set, where one local detector in the multiple local detectors corresponds to one local area of a type of target, and the type of target includes at least two local areas; detecting a to-be-recognized scene by using the multiple local detectors, and acquiring a feature, which is based on a local area of the target, of the to-be-recognized scene; and recognizing the to-be-recognized scene according to the feature, which is based on the local area of the target, of the to-be-recognized scene.

First claim

What is claimed is:

1. A scene recognition apparatus, comprising:
a non-transitory computer-readable medium having instructions stored thereon; and
a computer processor coupled to the non-transitory computer-readable medium and configured to execute the instructions to:
obtain multiple local detectors by training a training image set, wherein one local detector in the multiple local detectors corresponds to one local area of a type of target, and wherein the type of target comprises at least two local areas;
detect a to-be-recognized scene by using the multiple local detectors;
acquire a feature, which is based on a local area of the target, of the to-be-recognized scene, wherein the feature is acquired by selecting a maximum response value from a grid of multiple response values; and
recognize the to-be-recognized scene according to the feature, which is based on the local area of the target, of the to-be-recognized scene.

2. The apparatus according to claim 1 , wherein the computer processor is configured to execute the instruction to:
combine local detectors whose similarity is higher than a predetermined threshold in the multiple local detectors, to obtain a set in which the local detectors are combined;
detect the to-be-recognized scene by using the multiple local detectors;
acquire the feature, which is based on the local area of the target, of the to-be-recognized scene comprises detecting the to-be-recognized scene by using a local detector in the set in which the local detectors are combined; and
acquire the feature, which is based on the local area of the target, of the to-be-recognized scene.

3. The apparatus according to claim 2 , wherein the similarity comprises a similarity degree between features of local areas of training images corresponding to the multiple local detectors.

4. The apparatus according to claim 1 , wherein the computer processor is configured to execute the instructions to recognize the to-be-recognized scene according to the feature, which is based on the local area of the target, of the to-be-recognized scene comprises the computer processor being configured to execute the instructions to classify the feature, which is based on the local area of the target, of the to-be-recognized scene by using a classifier, and acquire a scene recognition result.

5. The apparatus according to claim 1 , wherein the computer processor is configured to execute the instructions to acquire the feature, which is based on the local area of the target, of the to-be-recognized scene comprises the computer processor being configured to:
acquire a response diagram of the to-be-recognized scene by using each local detector that detects the to-be-recognized scene;
divide the response diagram into multiple grids, using a maximum response value in each grid as a feature of each grid, using features of all grids of the response diagram as features corresponding to the response diagram; and
use features that correspond to the response diagram and are acquired by all local detectors that detect the to-be-recognized scene as the feature, which is based on the local area of the target, of the to-be-recognized scene.

6. A scene recognition apparatus, comprising:
a non-transitory computer-readable medium having instructions stored thereon; and
a computer processor coupled to the non-transitory computer-readable medium and configured to execute the instructions to:
obtain multiple local detectors by training a training image set, wherein one local detector in the multiple local detectors corresponds to one local area of a type of target, and wherein the type of target comprises at least two local areas;
detect a to-be-recognized scene by using the multiple local detectors;
acquire a feature, which is based on a local area of the target, of the to-be-recognized scene;
recognize the to-be-recognized scene according to the feature, which is based on the local area of the target, of the to-be-recognized scene;
combine local detectors whose similarity is higher than a predetermined threshold in the multiple local detectors, to obtain a set in which the local detectors are combined; and
detect the to-be-recognized scene by using a local detector in the set in which the local detectors are combined.

7. The apparatus according to claim 6 , wherein the similarity comprises a similarity degree between features of local areas of training images corresponding to the multiple local detectors.

8. The apparatus according to claim 6 , wherein the computer processor is configured to execute the instructions to classify the feature, which is based on the local area of the target, of the to-be-recognized scene by using a classifier, and acquire a scene recognition result.

9. The apparatus according to claim 6 , wherein the computer processor is configured to execute the instructions to:
acquire a response diagram of the to-be-recognized scene by using each local detector that detects the to-be-recognized scene;
divide the response diagram into multiple grids, use a maximum response value in each grid as a feature of each grid;
use features of all grids of the response diagram as features corresponding to the response diagram; and
use features that correspond to the response diagram and are acquired by all the local detectors that detect the to-be-recognized scene as the feature, which is based on the local area of the target, of the to-be-recognized scene.

10. A scene recognition apparatus, comprising:
a non-transitory computer-readable medium having instructions stored thereon; and
a computer processor coupled to the non-transitory computer-readable medium and configured to execute the instructions to:
obtain multiple local detectors by training a training image set, wherein one local detector in the multiple local detectors corresponds to one local area of a type of target, and wherein the type of target comprises at least two local areas;
detect a to-be-recognized scene by using the multiple local detectors;
acquire a feature, which is based on a local area of the target, of the to-be-recognized scene;
recognize the to-be-recognized scene according to the feature, which is based on the local area of the target, of the to-be-recognized scene;
acquire a response diagram of the to-be-recognized scene by using each local detector that detects the to-be-recognized scene;
divide the response diagram into multiple grids;
use a maximum response value in each grid as a feature of each grid;
use features of all the grids of the response diagram as features corresponding to the response diagram; and
use features that correspond to the response diagram and are acquired by all local detectors that detect the to-be-recognized scene as the feature, which is based on the local area of the target, of the to-be-recognized scene.

11. The apparatus according to claim 10 , wherein the response diagram is divided into a 1 by 1 grid, a 2 by 2 grid, and a 3 by 1 grid.

12. The apparatus according to claim 10 , wherein the computer processor is configured to:
execute the instruction to combine local detectors whose similarity is higher than a predetermined threshold in the multiple local detectors, to obtain a set in which the local detectors are combined;
execute the instructions to detect the to-be-recognized scene by using the multiple local detectors; and
acquire the feature, which is based on the local area of the target, of the to-be-recognized scene comprises detecting the to-be-recognized scene by using a local detector in the set in which the local detectors are combined.

13. The apparatus according to claim 12 , wherein the local detector

Assignees

• Huawei Tech Co Ltd

Inventors

• Jiang Yugang
• Liu Jie
• Wang Dong
• Zheng Yingbin
• Xue Xiangyang

Classifications

• G06V10/806

  of extracted features · CPC title

• G06V20/35 Primary

  Categorising the entire scene, e.g. birthday party or wedding scene · CPC title

• G06V20/00

  Scenes; Scene-specific elements (control of digital cameras H04N23/60) · CPC title

• G06F18/253

  of extracted features · CPC title

• G06K9/629

  Physics · mapped topic