Extracting salient features from video using a neurosynaptic system

• US 9,922,266 B2
• Filed: 03/31/2016
• Issued: 03/20/2018
• Est. Priority Date: 04/29/2014
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Andreopoulos, Alexander, Esser, Steven K., Modha, Dharmendra S.
• Primary Examiner(s): Carter, Aaron W
• Application Number:
  US15/087,945
• Publication Number:
  US 20170243076A1
• Time in Patent Office: 719 Days
• CPC Class Codes: G06N 3/049 Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
• CPC Class Codes: G06N 3/063 using electronic means
• CPC Class Codes: G06V 10/44 Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
• CPC Class Codes: G06V 10/462 Salient features, e.g. scale invariant feature transforms [SIFT]
• CPC Class Codes: G06V 10/464 using a plurality of salient features, e.g. bag-of-words [BoW] representations
• CPC Class Codes: G06V 10/56 relating to colour
• CPC Class Codes: G06V 20/46 Extracting features or characteristics from the video content, e.g. video fingerprints, representative shots or key frames
• CPC Class Codes: G06V 30/194 References adjustable by an adaptive method, e.g. learning

Abstract

Embodiments of the invention provide a method of visual saliency estimation comprising receiving an input sequence of image frames. Each image frame has one or more channels, and each channel has one or more pixels. The method further comprises, for each channel of each image frame, generating corresponding neural spiking data based on a pixel intensity of each pixel of the channel, generating a corresponding multi-scale data structure based on the corresponding neural spiking data, and extracting a corresponding map of features from the corresponding multi-scale data structure. The multi-scale data structure comprises one or more data layers, wherein each data layer represents a spike representation of pixel intensities of a channel at a corresponding scale. The method further comprises encoding each map of features extracted as neural spikes.

Claims

1. A method comprising:

receiving an input sequence of image frames; and

utilizing one or more neurosynaptic core circuits to perform operations including;

generating neural spiking data for the input sequence in accordance with a coding scheme, wherein the neural spiking data comprises, for each pixel of each image frame of the input sequence, neural spikes representing pixel intensity of the pixel, the neural spikes delivered to target axons, and a total number of the target axons that each of the neural spikes is delivered to is based on the coding scheme;

processing the neural spiking data to extract salient features; and

detecting at least one region of interest in the input sequence based on the salient features extracted.

2. The method of claim 1, wherein the salient features extracted include at least one motion saliency feature indicating at least one region of interest in the input sequence where motion is detected.

3. The method of claim 1, wherein the salient features extracted include at least one spatial saliency feature indicating at least one motionless region of interest in the input sequence.

4. The method of claim 1, further comprising:

encoding the salient features extracted as neural spikes.

5. The method of claim 1, further comprising:

for each region of interest detected, applying an object recognition algorithm to the region of interest detected to classify at least one object within the region of interest detected.

6. 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 instructions including;

receiving an input sequence of image frames; and

utilizing one or more neurosynaptic core circuits to perform operations including;

generating neural spiking data for the input sequence in accordance with a coding scheme, wherein the neural spiking data comprises, for each pixel of each image frame of the input sequence, neural spikes representing pixel intensity of the pixel, the neural spikes delivered to target axons, and a total number of the target axons that each of the neural spikes is delivered to is based on the coding scheme;

processing the neural spiking data to extract salient features; and

detecting at least one region of interest in the input sequence based on the salient features extracted.

7. The system of claim 6, wherein the salient features extracted include at least one motion saliency feature indicating at least one region of interest in the input sequence where motion is detected.

8. The system of claim 6, wherein the salient features extracted include at least one spatial saliency feature indicating at least one motionless region of interest in the input sequence.

9. The system of claim 6, wherein the instructions further include:

encoding the salient features extracted as neural spikes.

10. The system of claim 6, wherein the instructions further include:

for each region of interest detected, applying an object recognition algorithm to the region of interest detected to classify at least one object within the region of interest detected.

11. 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:

receiving an input sequence of image frames; and

utilizing one or more neurosynaptic core circuits to perform operations including;

generating neural spiking data for the input sequence in accordance with a coding scheme, wherein the neural spiking data comprises, for each pixel of each image frame of the input sequence, neural spikes representing pixel intensity of the pixel, the neural spikes delivered to target axons, and a total number of the target axons that each of the neural spikes is delivered to is based on the coding scheme;

processing the neural spiking data to extract salient features; and

detecting at least one region of interest in the input sequence based on the salient features extracted.

12. The non-transitory computer-readable medium of claim 11, wherein the salient features extracted include at least one motion saliency feature indicating at least one region of interest in the input sequence where motion is detected.

13. The non-transitory computer-readable medium of claim 11, wherein the salient features extracted include at least one spatial saliency feature indicating at least one motionless region of interest in the input sequence.

14. The non-transitory computer-readable medium of claim 11, wherein the operations further comprise:

encoding the salient features extracted as neural spikes.

15. The non-transitory computer-readable medium of claim 11, wherein the operations further comprise:

for each region of interest detected, applying an object recognition algorithm to the region of interest detected to classify at least one object within the region of interest detected.