Product auditing in point-of-sale images

1. A method comprising:

analyzing, by executing an instruction with a processor, a first image with an automatic image recognition process to identify products in the first image;

displaying, via a user interface, the first image with indications of the products identified in the first image by the automatic image recognition process;

subsequent to the analyzing of the first image with the automatic image recognition process, comparing, by executing an instruction with the processor, a region of interest of the first image displayed via the user interface with a plurality of reference product images stored in a database to identify a plurality of candidate product images from the plurality of reference product images as potential matches to a first product depicted in the region of interest, the candidate product images being associated with respective confidence levels indicating respective likelihoods of matching the first product;

displaying, via the user interface, the candidate product images simultaneously with the first image in a manner based on the respective confidence levels; and

automatically selecting, by executing an instruction with the processor, a first one of the candidate product images as matching the first product based on the respective confidence levels.

2. The method of claim 1 , wherein the comparing of the region of interest with the plurality of reference product images includes:

detecting first corners of the region of interest;

extracting first features from the detected first corners of the region of interest;

detecting second corners of a first reference product image of the plurality of reference product images;

extracting second features from the detected second corners;

determining relations between the first features and the second features; and

determining a dominant transformation projection between the region of interest and the first reference product image based on the relations.

3. The method of claim 2 , further including:

determining a number of inliers of the first features based on the dominant transformation projection; and

determining a first one of the respective confidence levels associated with the first reference product image based on the number of the inliers.

4. The method of claim 1 , wherein the displaying of the indications includes displaying the first image as a segmented image including one or more frames corresponding to respective products identified in the first image by the automatic image recognition process, the one or more frames including respective product names overlaying the respective products identified in the respective frames.

5. The method of claim 1 , wherein the candidate product images are displayed in order of the respective confidence levels.

6. The method of claim 1 , further including displaying values representative of the respective confidence levels adjacent to the candidate product images.

7. The method of claim 1 , further including displaying the first one of the candidate product images in a manner indicating the first one of the candidate product images has been selected as being matched to the first product.

8. An apparatus comprising:

a product database storing a plurality of reference product images;

an automatic image recognition application to identify products in a first image;

a candidate product searcher to:

subsequent to operation of the automatic image recognition application, compare a region of interest of the first image with the plurality of reference product images to identify a plurality of candidate product images from the plurality of reference product images as potential matches to a first product depicted in the first image, the plurality of candidate product images being associated with respective confidence levels indicating respective likelihoods of matching the first product; and

automatically select a first one of the candidate product images as matching the first product based on the respective confidence levels; and

a user interface to:

display the first image and one or more frames corresponding to respective products identified by the automatic image recognition application, the one or more frames including respective product names overlaying the respective products identified in the respective frames; and

display the candidate product images identified by the candidate product searcher simultaneously with the first image in a manner based on the respective confidence levels.

9. The apparatus of claim 8 , wherein, to compare the region of interest with the plurality of reference product images, the candidate product searcher is to:

detect first corners of the region of interest;

extract first features from the detected first corners of the region of interest;

detect second corners of a first reference product image of the plurality of reference product images;

extract second features from the detected second corners;

determine relations between the first features and the second features; and

determine a dominant transformation projection between the region of interest and the first reference product image based on the relations.

10. The apparatus of claim 9 , wherein the candidate product searcher is further to:

determine a number of inliers of the first features based on the dominant transformation projection; and

determine a first one of the respective confidence levels associated with the first reference product image based on the number of the inliers.

11. The apparatus of claim 8 , wherein, to compare the region of interest of the first image with the plurality of reference product images to identify the plurality of candidate product images, the candidate product searcher is to:

assign the respective confidence levels to the reference product images; and

compare the respective confidence levels associated with the reference product images to a threshold.

12. The apparatus of claim 8 , wherein the candidate product images are displayed in order of the respective confidence levels.

13. The apparatus of claim 8 , wherein the user interface is further to display values representative of the respective confidence levels adjacent to the candidate product images.

14. A tangible machine readable storage medium comprising instructions that, when executed, cause a machine to at least:

execute an automatic image recognition process to identify products in a first image;

display, via a user interface, the first image with indications of the products identified in the first image by the automatic image recognition process;

subsequent to execution of the automatic image recognition process, compare a region of interest of the first image with a plurality of reference product images stored in a database to identify a plurality of candidate product images from the plurality of reference product images as potential matches to a first product depicted in the first image, the candidate product images being associated with respective confidence levels indicating respective likelihoods of matching the first product;

display, via the user interface, the candidate product images simultaneously with the first image in a manner based on the respective confidence levels; and

automatically select a first one of the candidate product images as matching the first product based on the respective confidence levels.

15. The tangible machine readable storage medium of claim 14 , wherein, to compare the region of interest with the plurality of reference product images, the instructions, when executed, cause the machine to:

detect first corners of the region of interest;

extract first features from the detected first corners of the region of interest;

detect second corners of a first reference product image of the plurality of reference product images;

extract second features from the detected second corners;

determine relations between the first features and the second features; and

determine a dominant transformation projection between the region of interest and the first reference product image based on the relations.

16. The tangible machine readable storage medium of claim 15 , wherein the instructions, when executed, further cause the machine to:

determine a number of inliers of the first features based on the dominant transformation projection; and

determine a first one of the respective confidence levels associated with the first reference product image based on the number of the inliers.

17. The tangible machine readable storage medium of claim 14 , wherein the instructions, when executed, further cause the machine to at least one of create a record or update the record to indicate the first product matched the first one of the candidate product images.

18. The tangible machine readable storage medium of claim 14 , wherein the candidate product images are displayed in order of the respective confidence levels.

19. The tangible machine readable storage medium of claim 14 , wherein the instructions, when executed, further cause the machine to display, via the user interface, values representative of the respective confidence levels adjacent to the candidate product images.

20. The tangible machine readable storage medium of claim 14 , wherein the instructions, when executed, further cause the machine to display, via the user interface, the first one of the candidate product images in a manner indicating the first one of the candidate product images has been selected as being matched to the first product.