Apparatus and method for object recognition

Patent metadata

Field	Value
Publication number	US-11461920-B2
Application number	US-202016879215-A
Country	US
Kind code	B2
Filing date	May 20, 2020
Priority date	Feb 2, 2018
Publication date	Oct 4, 2022
Grant date	Oct 4, 2022

Abstract

A system and method for recognizing an object. The system includes an imaging apparatus for capturing an image of an object and a processor for receiving the captured image of the object, and for, when it is determined that fewer than a predetermined number of objects have been previously imaged, for determining whether the image of the captured object includes one or more characteristics determined to be similar to a same characteristic in a group of previously imaged objects so that the captured image is grouped with the previously imaged objects, or whether the image of the captured object includes one or more characteristics determined to be dissimilar to a same characteristic in a group of previously imaged objects so that the captured image is not grouped with the previously imaged objects, and the image of the captured object starts another group of previously imaged objects.

First claim

What is claimed:

1. A system, comprising
an imaging apparatus;
a display;
a processor; and
a memory storing instructions that, when executed by the processor, cause the processor to perform operations comprising:
receiving a captured image of an object from the imaging apparatus;
deriving a similarity between one or more characteristics of the captured image and one or more corresponding characteristics of a group of images based on a similarity function, wherein the group of images comprises images of previously imaged objects;
determining that the similarity satisfies a similarity condition;
in response to determining that the similarity satisfies the similarity condition, adding the captured image of the object to the group of images;
determining that a measure of variation in a visual characteristic among the group of images satisfies a condition; and
in response to determining that the measure of variation satisfies the condition, training a supervised or unsupervised visual learning system to recognize a type of the object based on the group of images.

2. The system of claim 1 , wherein the object comprises a medication pill.

3. The system of claim 1 , wherein the object comprises a medication liquid.

4. The system of claim 1 , wherein the object comprises a medical device.

5. The system of claim 1 , wherein the operations comprise
causing the imaging apparatus to capture one or more further images of an identifier on a vessel associated with the object, wherein the identifier is associated with the group of images.

6. The system of claim 5 , wherein the operations comprise retrieving one or more images associated with the identifier from a remote data storage location.

7. The system of claim 1 , wherein the measure of variation in the visual characteristic comprises a measure of variation in lighting.

8. The system of claim 1 , wherein the one or more characteristics comprise one or more of color, shape, surface reflectivity, or surface markings.

9. The system of claim 8 , wherein the one or more characteristics comprise color and at least one other characteristic, and wherein deriving the similarity comprises weighting color more heavily than at least one other characteristic when deriving the similarity.

10. The system of claim 1 , wherein the condition comprises that a number of scenarios for the visual characteristic that are portrayed in the group of images is at least a threshold value.

11. A method of training a system to recognize an object, comprising:
presenting a first object to an imaging system;
capturing an image of the first object by the imaging system;
storing the image of the first object in a first group of images;
determining that a measure of variation in a visual characteristic among the first group of images does not satisfy a condition;
until it is determined that the measure of variation satisfies the condition,
obtaining an image of an additional object,
testing whether the image of the additional object satisfies a similarity condition with images in an existing group of images, wherein the similarity condition is based on one or more dimensions of image characteristics,
obtaining, as a result of the testing of the image of the additional object, a determination of whether the image of the additional object satisfies the similarity condition with images in an existing group of images, and
in response to determining that the image of the additional object satisfies the similarity condition with images in the existing group of images, grouping the image of the additional object into the existing group of images for which the similarity condition is satisfied, or
in response to determining that the image of the additional object does not satisfy the similarity condition with images in any existing group of images, grouping the image of the additional object into a new group of images distinct from existing groups of images;

determining that the measure of variation satisfies the condition; and
in response to determining that the measure of variation satisfies the condition, training a supervised or unsupervised visual learning system to recognize a type of the first object based on the first group of images.

12. The method of claim 11 , further comprising, subsequent to training the supervised or unsupervised visual learning system,
presenting a third object to the imaging system;
capturing an image of the third object by the imaging system; and
determining, by the supervised or unsupervised visual learning system, whether a type of the third object is the type of the first object.

13. The method of claim 11 , wherein the first object is a medication pill.

14. The method of claim 11 , wherein the first object is a medication liquid.

15. The method of claim 11 , wherein the first object is a medical device.

16. The method of claim 11 , wherein the measure of variation in the visual characteristic comprises a measure of variation in lighting.

17. The method of claim 11 , wherein the one or more dimensions comprise one or more of color, shape, surface reflectivity, or surface markings.

18. The method of claim 17 , wherein the one or more dimensions comprise color and at least one other dimension, and wherein color is weighted more heavily than the at least one other dimension for testing of whether the similarity condition is satisfied.

19. The method of claim 11 , wherein the condition comprises that a number of scenarios for the visual characteristic that are portrayed in the first group of images is at least a threshold value.

Assignees

• Aic Innovations Group Inc

Inventors

• Glasner Daniel
• Guan Lei
• Hanina Adam
• Zhang Li

Classifications

• G06V30/19173

  Classification techniques · CPC title

• G06V10/82 Primary

  using neural networks · CPC title

• G06V10/761

  Proximity, similarity or dissimilarity measures · CPC title

• G06T7/70 Primary

  Determining position or orientation of objects or cameras (camera calibration G06T7/80) · CPC title

• G06F18/24133

  Distances to prototypes · CPC title