METHOD AND APPARATUS FOR RECOGNIZING OBJECT, AND METHOD AND APPARATUS FOR TRAINING RECOGNITION MODEL

• US 20170140247A1
• Filed: 11/09/2016
• Published: 05/18/2017
• Est. Priority Date: 11/16/2015
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: YOO, Byungin, KWAK, Youngjun, HWANG, Wonjun, KIM, Jungbae, HAN, Seungju, HAN, Jaejoon, RHEE, Seon Min
• Granted Patent: US 10,860,887 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 18/214 Generating training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06N 3/0499 Feedforward networks
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06N 3/09 Supervised learning
• CPC Class Codes: G06V 10/451 with interaction between the filter responses, e.g. cortical complex cells
• CPC Class Codes: G06V 10/82 using neural networks
• CPC Class Codes: G06V 40/1347 Preprocessing; Feature extraction
• CPC Class Codes: G06V 40/1365 Matching; Classification
• CPC Class Codes: G06V 40/16 Human faces, e.g. facial parts, sketches or expressions
• CPC Class Codes: G06V 40/168 Feature extraction; Face representation
• CPC Class Codes: G06V 40/172 Classification, e.g. identification
• CPC Class Codes: H04L 63/0861 using biometrical features, e.g. fingerprint, retina-scan cryptographic mechanisms or cryptographic arrangements for entity authentication using biological data H04L9/3231

Abstract

A method and an apparatus for recognizing an object are disclosed. The apparatus may extract a plurality of features from an input image using a single recognition model and recognize an object in the input image based on the extracted features. The single recognition model may include at least one compression layer configured to compress input information and at least one decompression layer configured to decompress the compressed information to determine the features.

Claims

1. A method of recognizing an object, the method comprising:

extracting a plurality of features from an input image using a single recognition model; and

recognizing an object in the input image based on the extracted features.

2. The method of claim 1, wherein the single recognition model comprises:

at least one compression layer configured to compress input information of the input image; and

at least one decompression layer configured to decompress the compressed information to determine the extracted features.

3. The method of claim 1, wherein the extracting of the features comprises:

determining a plurality of areas in the input image;

inputting information on the areas to the single recognition model; and

determining respective features of the areas using the single recognition model based on the input information.

4. The method of claim 1, wherein the single recognition model comprises:

a single input layer; and

a plurality of output layers configured to output the extracted features.

5. The method of claim 4, further comprising:

receiving information on the input image at the single input layer.

6. The method of claim 4, wherein the receiving receives information on a plurality of areas in the input image at the single input layer.

7. The method of claim 1, wherein the single recognition model comprises:

a plurality of input layers; and

a plurality of output layers configured to output the extracted features.

8. The method of claim 7, further comprising:

receiving information on a plurality of areas in the input image at the input layers.

9. The method of claim 3, wherein the extracting comprises:

compressing information of correlated areas among the areas at a first compressing layer of the single recognition model; and

compressing information on an entirety of the areas based on information transferred from the at least one first compression layer.

10. The method of claim 3, wherein the recognizing of the object comprises:

recognizing the object based on features extracted from the areas of the extracted features.

11. The method of claim 3, wherein the recognizing of the object comprises:

determining a probability of a presence of an occlusion in a current area among the areas using the single recognition model; and

applying a weight to a feature of the current area, the weight being based on the determined probability.

12. The method of claim 3, wherein the extracting comprises:

determining the features from the information on each of the areas using a plurality of layers in the single recognition model,

wherein at least one of the layers is connected to another layer of the layers.

13. The method of claim 1, wherein the extracting of the features comprises:

extracting the features from a determined area in the input image using the single recognition model.

14. The method of claim 1, wherein the single recognition model is a model trained in advance based on recognition results of respective recognition models associated with the features of the input image.

15. The method of claim 1, wherein the input image is a face image or a fingerprint image.

16. A computer program embodied on a non-transitory computer readable medium, when executed by a processor, configured to cause the processor to perform the method of claim 1.

17. An apparatus for recognizing an object, the apparatus comprising:

a memory storing computer-executable instructions; and

at least one processor configured to execute the instructions such that the processor is configured to,

extract a plurality of features from an input image using a single recognition model; and

recognize an object in the input image based on the extracted features.

18. The apparatus of claim 17, wherein the single recognition model comprises:

at least one compression layer configured to compress input information of the input image; and

at least one decompression layer configured to decompress the compressed information to determine the extracted features.

19. The apparatus of claim 17, wherein the apparatus image includes the single recognition model and the single recognition model comprises:

a single input layer; and

a plurality of output layers configured to output the extracted features, the single input layer coupled to the plurality of output layers.

20. The apparatus of claim 19, wherein the single input layer is configured to receive the input image.

21. The apparatus of claim 19, wherein the single input layer is configured to receive information on a plurality of areas in the input image.

22. The apparatus of claim 17, wherein the apparatus includes the single recognition model and the single recognition model comprises:

a plurality of input layers; and

a plurality of output layers configured to output the extracted features, the plurality of input layers coupled to the output layers.

23. The apparatus of claim 22, wherein the input layers are configured to receive information on the input image.

24. The apparatus of claim 17, wherein the at least one processor is configured to execute the computer-executable instructions to,

determine a plurality of areas in the input image;

and

input information on the areas to the single recognition model, and

determine respective features of the areas using the single recognition model.

25. The apparatus of claim 24, wherein the apparatus includes the single recognition model and the single recognition model comprises:

at least one first compression layer configured to compress information of correlated areas among the areas; and

a second compression layer configured to compress information on an entirety of the areas based on information transferred from the at least one first compression layer.