Scene understanding using a neurosynaptic system

• US 10,558,892 B2
• Filed: 09/28/2018
• Issued: 02/11/2020
• Est. Priority Date: 05/29/2014
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Andreopoulos, Alexander, Appuswamy, Rathinakumar, Datta, Pallab, Esser, Steven K., Modha, Dharmendra S.
• Primary Examiner(s): Ahmed, Samir A
• Application Number:
  US16/147,106
• Publication Number:
  US 20190042886A1
• Time in Patent Office: 501 Days
• CPC Class Codes: G06F 18/214 Generating training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06F 18/24 Classification techniques
• CPC Class Codes: G06N 3/065 Analogue means
• CPC Class Codes: G06N 3/08 Learning methods
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/20081 Training; Learning
• CPC Class Codes: G06T 7/246 using feature-based methods, e.g. the tracking of corners or segments
• CPC Class Codes: G06V 10/451 with interaction between the filter responses, e.g. cortical complex cells
• CPC Class Codes: G06V 10/955 using specific electronic processors
• CPC Class Codes: G06V 20/41 Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items segmenting video sequences G06V20/49
• CPC Class Codes: G06V 30/194 References adjustable by an adaptive method, e.g. learning
• CPC Class Codes: H04N 19/136 Incoming video signal characteristics or properties
• CPC Class Codes: H04N 9/67 for matrixing camera processing pipelines for matrixing of colour signals H04N23/85

Abstract

Embodiments of the invention provide a method for scene understanding based on a sequence of image frames. The method comprises converting each pixel of each image frame to neural spikes, and extracting features from the sequence of image frames by processing neural spikes corresponding to pixels of the sequence of image frames. The method further comprises encoding the extracted features as neural spikes, and classifying the extracted features.

Claims

1. A method comprising:

for an image frame of a sequence of image frames;

extracting, utilizing a neurosynaptic core circuit comprising a plurality of electronic synapse devices interconnecting a plurality of electronic neurons to a plurality of electronic axons, one or more features from the image frame;

determining a quantity and a frequency of neural spikes to generate by mapping the one or more features extracted to a number of neural spikes to generate; and

generating one or more neural spikes based on the quantity and the frequency determined, wherein the one or more neural spikes encode the one or more features extracted.

2. The method of claim 1, wherein the mapping comprises:

mapping a level of each of the one or more of the features extracted to the number of neural spikes to generate.

3. The method of claim 2, wherein the mapping is linear.

4. The method of claim 2, wherein the mapping is non-linear.

5. The method of claim 2, wherein the level represents a pixel intensity value.

6. The method of claim 2, further comprising:

determining the number of neural spikes to generate within a pre-specified time window based on the mapping.

7. The method of claim 1, wherein the one or more features extracted comprise one or more salient features.

8. The method of claim 1, wherein the one or more features extracted comprise one or more discriminative features.

9. A system comprising:

at least one processor; and

a non-transitory processor-readable memory device storing instructions that when executed by the at least one processor causes the at least one processor to perform operations including;

for an image frame of a sequence of image frames;

extracting, utilizing a neurosynaptic core circuit comprising a plurality of electronic synapse devices interconnecting a plurality of electronic neurons to a plurality of electronic axons, one or more features from the image frame;

determining a quantity and a frequency of neural spikes to generate by mapping the one or more features extracted to a number of neural spikes to generate; and

generating one or more neural spikes based on the quantity and the frequency determined, wherein the one or more neural spikes encode the one or more features extracted.

10. The system of claim 9, wherein the mapping comprises:

mapping a level of each of the one or more of the features extracted to the number of neural spikes to generate.

11. The system of claim 10, wherein the mapping is linear.

12. The system of claim 10, wherein the mapping is non-linear.

13. The system of claim 10, wherein the level represents a pixel intensity value.

14. The system of claim 10, wherein the operations further comprise:

determining the number of neural spikes to generate within a pre-specified time window based on the mapping.

15. The system of claim 9, wherein the one or more features extracted comprise one or more salient features.

16. The system of claim 9, wherein the one or more features extracted comprise one or more discriminative features.

17. A non-transitory computer-readable medium storing instructions that, when executed by at least one processor of a machine, cause the machine to perform operations comprising:

for an image frame of a sequence of image frames;

extracting, utilizing a neurosynaptic core circuit comprising a plurality of electronic synapse devices interconnecting a plurality of electronic neurons to a plurality of electronic axons, one or more features from the image frame;

determining a quantity and a frequency of neural spikes to generate by mapping the one or more features extracted to a number of neural spikes to generate; and

generating one or more neural spikes based on the quantity and the frequency determined, wherein the one or more neural spikes encode the one or more features extracted.