Realogram scene analysis of images: superpixel scene analysis
The techniques include an image recognition system to receive a realogram image including a plurality of organized objects and to detect and identify objects in the realogram image of one or more items on a retail shelf, identify shelf fronts and labels on the shelf fronts, identify empty space under shelves, identify areas where unidentified products may be, and identify areas where products are “out of stock”.
1. A computer-implemented method comprising:
receiving an image of a plurality of organized objects;
generating superpixels based on groups of similar pixels in the image of the plurality of organized objects;
labeling superpixels with information and observations from image analysis on the image of the plurality of organized objects, wherein labeling includes classifying superpixels as empty or other;
identifying empty areas under a shelf based on the classified superpixels;
iterating a segmentation algorithm on the classified superpixels to reclassify each of the superpixels as empty or other; and
identifying regions of the image of the plurality of organized objects, including empty shelf space, based on the reclassified superpixels.
2. The computer implemented method of claim 1 , further comprising identifying indexed objects in the image, wherein labeling superpixels with information and observations includes labeling superpixels with an object identifier.
3. The computer implemented method of claim 1 , further comprising generating region hypotheses using the information and observations.
4. The computer implemented method of claim 3 , further comprising validating the hypotheses.
5. The computer implemented method of claim 4 , further comprising classifying regions of the image using the validated hypotheses.
6. The computer implemented method of claim 1 , wherein the regions include one or more from the group of: known objects, empty shelf space, shelf features, and labels.
7. A system comprising:
one or more processors; and
a memory, the memory storing instructions, which when executed cause the one or more processors to:
receive an image of a plurality of organized objects;
generate superpixels based on groups of similar pixels in the image of the plurality of organized objects;
label superpixels with information and observations from image analysis on the image of the plurality of organized objects, wherein labeling includes classifying superpixels as empty or other;
identify empty areas under a shelf based on the classified superpixels;
iterate a segmentation algorithm on the classified superpixels to reclassify each of the superpixels as empty or other; and
identify regions of the image of the plurality of organized objects, including empty shelf space, based on the reclassified superpixels.
8. The system of claim 7 , wherein the instructions further cause the one or more processors to identify indexed objects in the image and label superpixels with an object identifier.
9. The system of claim 7 , wherein the instructions further cause the one or more processors to generate region hypotheses using the information and observations.
10. The system of claim 9 , wherein the instructions further cause the one or more processors to validate the hypotheses.
11. The system of claim 10 , wherein the instructions further cause the one or more processors to classify regions of the image using the validated hypotheses.
12. The system of claim 7 , wherein the regions include one or more from the group of: known objects, empty shelf space, shelf features, and labels.
13. A computer program product comprising a non-transitory computer usable medium including a computer readable program, wherein the computer readable program, when executed on a computer causes the computer to:
receive an image of a plurality of organized objects;
generate superpixels based on groups of similar pixels in the image of the plurality of organized objects;
label superpixels with information and observations from image analysis on the image of the plurality of organized objects, wherein labeling includes classifying superpixels as empty or other;
identify empty areas under a shelf based on the classified superpixels;
iterate a segmentation algorithm on the classified superpixels to reclassify each of the superpixels as empty or other; and
identify regions of the image of the plurality of organized objects, including empty shelf space, based on the reclassified superpixels.
14. The computer program product of claim 13 , wherein the computer readable program further causes the computer to identify indexed objects in the image and label superpixels with an object identifier.
15. The computer program product of claim 13 , wherein the computer readable program further causes the computer to generate region hypotheses using the information and observations.
16. The computer program product of claim 15 , wherein the computer readable program further causes the computer to classify regions of the image using the region hypotheses.
17. The computer program product of claim 13 , wherein the regions include one or more from the group of: known objects, empty shelf space, shelf features, and labels.