Optimizations for structure mapping and up-sampling

Patent metadata

Field	Value
Publication number	US-10796452-B2
Application number	US-201816236877-A
Country	US
Kind code	B2
Filing date	Dec 31, 2018
Priority date	Dec 3, 2017
Publication date	Oct 6, 2020
Grant date	Oct 6, 2020

Abstract

In one embodiment, a system accesses a probability model associated with an image depicting a body. The probability model includes probability values associated with regions of the image and each probability value represents a probability of the associated region of the image containing a particular body part. The system selects a subset (e.g., 3) of the probability values based on a comparison of the probability values. For each selected probability value, the system identifies surrounding probability values surrounding the selected probability value and computes a probabilistic maximum based on the selected probability value and the surrounding probability values. Each probabilistic maximum is associated with a location within the regions associated with the selected probability value and the surrounding probability values. One of the locations associated with the probabilistic maxima is then selected, which represents a determined location in the image that corresponds to the particular body part in the image.

First claim

What is claimed is:

1. A method comprising, by a computing system:
accessing a probability model associated with an image depicting a body, wherein the probability model comprises probability values associated with regions of the image, wherein each of the probability values represents a probability of the associated region of the image containing a predetermined body part;
selecting a subset of the probability values based on a comparison of the probability values;
for each selected probability value in the selected subset of probability values:
identifying surrounding probability values whose associated regions surround the region associated with the selected probability value; and
computing a probabilistic maximum based on the selected probability value and the surrounding probability values, the probabilistic maximum being associated with a location within the regions associated with the selected probability value and the surrounding probability values; and

selecting, based on the probabilistic maxima, one of the locations associated with the probabilistic maxima, the selected location representing a determined location in the image that corresponds to the predetermined body part of the body.

2. The method of claim 1 , wherein each probability value in the subset of probability values is a local maximum.

3. The method of claim 1 , wherein for each selected probability value in the selected subset of probability values, the region associated with that selected probability value and the regions associated with the identified surrounding probability values form a 3×3 region, a 5×5 region, or a 7×7 region.

4. The method of claim 1 , wherein the probabilistic maximum is computed using quadrilateral interpolation.

5. The method of claim 1 , wherein each of the locations associated with the probabilistic maxima is a point.

6. The method of claim 1 , wherein the probabilistic maximum associated with the selected location is greater than or equal to any of the other probabilistic maxima.

7. The method of claim 1 , wherein the regions associated with the probability values are rectangular and aligned horizontally and vertically.

8. The method of claim 1 , wherein a number of probability values in the subset of probability values is less than one-hundredth of a number of probability values in the probability model.

9. A system comprising: one or more processors and one or more computer-readable non-transitory storage media coupled to one or more of the processors, the one or more computer-readable non-transitory storage media comprising instructions operable when executed by one or more of the processors to cause the system to perform operations comprising:
access a probability model associated with an image depicting a body, wherein the probability model comprises probability values associated with regions of the image, wherein each of the probability values represents a probability of the associated region of the image containing a predetermined body part;
select a subset of the probability values based on a comparison of the probability values;
for each selected probability value in the selected subset of probability values:
identify surrounding probability values whose associated regions surround the region associated with the selected probability value; and
compute a probabilistic maximum based on the selected probability value and the surrounding probability values, the probabilistic maximum being associated with a location within the regions associated with the selected probability value and the surrounding probability values; and

select, based on the probabilistic maxima, one of the locations associated with the probabilistic maxima, the selected location representing a determined location in the image that corresponds to the predetermined body part of the body.

10. The system of claim 9 , wherein each probability value in the subset of probability values is a local maximum.

11. The system of claim 9 , wherein for each selected probability value in the selected subset of probability values, the region associated with that selected probability value and the regions associated with the identified surrounding probability values form a 3×3 region, a 5×5 region, or a 7×7 region.

12. The system of claim 9 , wherein the probabilistic maximum is computed using quadrilateral interpolation.

13. The system of claim 9 , wherein each of the locations associated with the probabilistic maxima is a point.

14. The system of claim 9 , wherein the probabilistic maximum associated with the selected location is greater than or equal to any of the other probabilistic maxima.

15. One or more computer-readable non-transitory storage media embodying software that is operable when executed to cause one or more processors to perform operations comprising:
access a probability model associated with an image depicting a body, wherein the probability model comprises probability values associated with regions of the image, wherein each of the probability values represents a probability of the associated region of the image containing a predetermined body part;
select a subset of the probability values based on a comparison of the probability values;
for each selected probability value in the selected subset of probability values:
identify surrounding probability values whose associated regions surround the region associated with the selected probability value; and
compute a probabilistic maximum based on the selected probability value and the surrounding probability values, the probabilistic maximum being associated with a location within the regions associated with the selected probability value and the surrounding probability values; and

select, based on the probabilistic maxima, one of the locations associated with the probabilistic maxima, the selected location representing a determined location in the image that corresponds to the predetermined body part of the body.

16. The media of claim 15 , wherein each probability value in the subset of probability values is a local maximum.

17. The media of claim 15 , wherein for each selected probability value in the selected subset of probability values, the region associated with that selected probability value and the regions associated with the identified surrounding probability values form a 3×3 region, a 5×5 region, or a 7×7 region.

18. The media of claim 15 , wherein the probabilistic maximum is computed using quadrilateral interpolation.

19. The media of claim 15 , wherein each of the locations associated with the probabilistic maxima is a point.

20. The media of claim 15 , wherein the probabilistic maximum associated with the selected location is greater than or equal to any of the other probabilistic maxima.

Assignees

• Facebook Inc

Inventors

• Vajda Peter
• Zhang Peizhao
• Uyttendaele Matthieu Tony
• Wang Yanghan

Classifications

• G06N3/084 Primary

  Backpropagation, e.g. using gradient descent · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/0495

  Quantised networks; Sparse networks; Compressed networks · CPC title