Methods and apparatus to perform image analyses in a computing environment
An example apparatus includes a feature extractor to generate a first image descriptor based on a first image of a first retail product tag corresponding to a first category, the first image descriptor representative of one or more visual features of the first retail product tag; a feature descriptor generator to generate a feature descriptor corresponding to the first retail product tag by concatenating the first image descriptor and a first category signature corresponding to the first retailer category; and a classifier to generate a first probability value corresponding to a first type of promotional product tag and a second probability value corresponding to a second type of promotional product tag based on the feature descriptor; and determine whether the first retail product tag corresponds to the first type of promotional product tag or the second type of promotional product tag based on the first and second probability values.
1. An apparatus to classify a first retail product tag image, the apparatus comprising:
memory;
instructions; and
at least one processor to execute the instructions to:
determine, based on operations, a first visual feature of the first retail product tag image, the first retail product tag image corresponding to a first retailer category;
generate a feature descriptor corresponding to the first retail product tag image based on (a) a first image descriptor, the first image descriptor indicative of the first visual feature, and (b) a category signature, the category signature indicative of the first retail category, the category signature based on samples of a probability distribution function (PDF) of second image descriptors corresponding to the first retailer category;
generate a first probability value associated with a first type of promotional product tag and a second probability value associated with a second type of promotional product tag, the second type of promotional product tag different than the first type of promotional product tag; and
determine a respective one of the first type of promotional product tag or the second type of promotional product tag of the first retail product tag image based on the first probability value and second probability value to differentiate the first retail product tag image corresponding to the first retailer category from a second retail product tag image corresponding to a second retailer category.
2. The apparatus of claim 1 , wherein the operations include a convolution operation and a pooling operation to determine the first visual feature.
3. The apparatus of claim 1 , wherein the first image descriptor is based on the first visual feature and a second visual feature of the first retail product tag image.
4. The apparatus of claim 1 , wherein the at least one processor is to generate the feature descriptor by concatenating the category signature with the first image descriptor.
5. The apparatus of claim 4 , wherein the at least one processor is to generate the feature descriptor, the feature descriptor having a bit length at least twice as long as the first image descriptor.
6. The apparatus of claim 1 , wherein the probability distribution function (PDF) is a Gaussian probability distribution function (PDF), and wherein the at least one processor is to sample the Gaussian probability distribution function (PDF) of the second image descriptors to generate the category signature, the second image descriptors including the first image descriptor.
7. The apparatus of claim 1 , wherein the first type of promotional product tag is a multi-buy product tag and the second type of promotional product tag is a price reduction product tag.
8. A non-transitory computer-readable storage medium comprising instructions that, when executed, cause a processor to at least:
determine, based on operations, a first visual feature of a first retail product tag image, the first retail product tag image corresponding to a first retailer category;
generate a feature descriptor corresponding to the first retail product tag image based on (a) a first image descriptor, the first image descriptor indicative of the first visual feature, and (b) a category signature, the category signature indicative of the first retail category, the category signature being based on a probability distribution function (PDF) of second image descriptors corresponding to the first retailer category;
generate a first probability value associated with a first type of promotional product tag and a second probability value associated with a second type of promotional product tag, the second type of promotional product tag different than the first type of promotional product tag; and
determine a respective one of the first type of promotional product tag or the second type of promotional product tag of the first retail product tag image based on the first probability value and second probability value to differentiate the first retail product tag image corresponding to the first retailer category from a second retail product tag image corresponding to a second retailer category.
9. The non-transitory computer-readable storage medium of claim 8 , wherein the operations include a convolution operation and a pooling operation to determine the first visual feature.
10. The non-transitory computer-readable storage medium of claim 8 , wherein the first image descriptor is based on the first visual feature and a second visual feature of the first retail product tag image.
11. The non-transitory computer-readable storage medium of claim 8 , including further instructions that, when executed, cause the processor to generate the feature descriptor by concatenating the category signature with the first image descriptor.
12. The non-transitory computer-readable storage medium of claim 11 , wherein the instructions, when executed, cause the processor to generate the feature descriptor, the feature descriptor having a bit length at least twice as long as the first image descriptor.
13. The non-transitory computer-readable storage medium of claim 8 , wherein the probability distribution function (PDF) is a Gaussian probability distribution function (PDF), and including further instructions that, when executed, cause the processor to sample the Gaussian probability distribution function (PDF) of the second image descriptors to generate the category signature, the second image descriptors including the first image descriptor.
14. The non-transitory computer-readable storage medium of claim 8 , wherein the first type of promotional product tag is a multi-buy product tag and the second type of promotional product tag is a price reduction product tag.
15. A method to classify a first retail product tag image, the method comprising:
determining, based on operations, a first visual feature of the first retail product tag image, the first retail product tag image corresponding to a first retailer category;
generating a feature descriptor corresponding to the first retail product tag image based on (a) a first image descriptor, the first image descriptor indicative of the first visual feature, and (b) a category signature, the category signature indicative of the first retail category, the category signature being based on samples of a probability distribution function (PDF) of a plurality of image descriptors corresponding to the first retailer category;
generating a first probability value associated with a first type of promotional product tag and a second probability value associated with a second type of promotional product tag, the second type of promotional product tag different than the first type of promotional product tag; and
determining a respective one of the first type of promotional product tag or the second type of promotional product tag of the first retail product tag image based on the first probability value and second probability value to differentiate the first retail product tag image corresponding to the first retailer category from a second retail product tag image corresponding to a second retailer category.
16. The method of claim 15 , wherein the operations include a convolution operation and a pooling operation to determine the first visual feature.
17. The method of claim 15 , wherein the first image descriptor is based on the first visual feature and a second visual feature of the first retail product tag image.
18. The method of claim 15 , further including generating the feature descriptor by concatenating the category signature with the first image descriptor.
19. The method of claim 18 , further including generating the feature descriptor, the feature descriptor having a bit length at least twice as long as the first image descriptor.
20. The method of claim 15 , further including generating the category signature by sampling the probability distribution function (PDF) of the plurality of image descriptors corresponding to the first retailer category, the plurality of image descriptors including the first image descriptor, the probability distribution function (PDF) being a Gaussian probability distribution function (PDF).