Logo recognition in images and videos
Methods, apparatus, systems and articles of manufacture of logo recognition in images and videos are disclosed. An example method to detect a specific brand in images and video streams comprises accepting luminance images at a scale in an x direction Sx and a different scale in a y direction Sy in a neural network, and training the neural network with a set of training images for detected features associated with a specific brand.
1. A method to detect a specific brand in images and video streams, comprising:
accepting luminance images at a scale in an x direction Sx and a different scale in a y direction Sy in a neural network; and
training the neural network with a set of training images for detected features associated with a specific brand.
2. The method as defined in claim 1 , further including accepting RGB images at a scale in the x direction Sx and a different scale in the y direction Sy in a neural network.
3. The method as defined in claim 1 , further including learning to classify an image into the specific brand by applying a logo image for the specific brand in the set of training images.
4. The method as defined in claim 1 , further including generating a set of training images with a set of logo images associated with the specific brand and located on backgrounds associated with the specific brand.
5. The method as defined in claim 1 , further including segmenting a detected character image using heuristics for stroke width, number of horizontal strokes, number of vertical strokes, and number of loops.
6. The method as defined in claim 1 , further including identifying contours according to vertical stroke density and transition analysis.
7. The method as defined in claim 1 , further including:
combining neighboring keypoint regions with consistent aspect ratios and size to generate a new keypoint region;
detecting and combining edge segments of an image in a keypoint region; and
binning sample points on selected edges by angle and distance with reference to a dominant orientation.
8. An apparatus, comprising:
at least one processor; and
a memory in communication with the at least one processor, the memory including non-transitory computer-readable code which, when executed, causes the at least one processor to at least:
accept luminance images at a scale in an x direction Sx and a different scale in a y direction Sy in a neural network; and
train the neural network with a set of training images for detected features associated with a specific brand.
9. The apparatus as defined in claim 8 , wherein the code, when executed, causes the at least one processor to accept RGB images at a scale in the x direction Sx and a different scale in the y direction Sy in the neural network.
10. The apparatus as defined in claim 8 , wherein the code, when executed, causes the at least one processor to learn to classify an image into the specific brand by applying a logo image for the specific brand in the set of training images.
11. The apparatus as defined in claim 8 , wherein the code, when executed, causes the at least one processor to generate a set of training images associated with the specific brand and located on backgrounds associated with the specific brand.
12. The apparatus as defined in claim 8 , wherein the code, when executed, causes the at least one processor to segment a detected character image using heuristics for stroke width, number of horizontal strokes, number of vertical strokes, and number of loops.
13. The apparatus as defined in claim 8 , wherein the code, when executed, causes the at least one processor to identify contours according to vertical stroke density and transition analysis.
14. The apparatus as defined in claim 8 , wherein the code, when executed, causes the at least one processor to:
combine neighboring keypoint regions with consistent aspect ratios and size to generate a new keypoint region;
detect and combine edge segments of an image in a keypoint region; and
bin sample points on selected edges by angle and distance with reference to a dominant orientation.
15. A non-transitory computer readable storage medium comprising code that, when executed, causes a machine to at least:
accept luminance images at a scale in an x direction Sx and a different scale in a y direction Sy in a neural network; and
train the neural network with a set of training images for detected features associated with a specific brand.
16. The storage medium as defined in claim 15 , wherein the code, when executed, causes the machine to learn to classify an image into the specific brand by applying a logo image for the specific brand in the set of training images.
17. The storage medium as defined in claim 15 , wherein the code, when executed, causes the machine to generate a set of training images with a set of logo images associated with the specific brand and located on backgrounds associated with the specific brand.
18. The storage medium as defined in claim 15 , wherein the code, when executed, causes the machine to segment a detected character image using heuristics for stroke width, number of horizontal strokes, number of vertical strokes, and number of loops.
19. The storage medium as defined in claim 15 , wherein the code, when executed causes the machine to identify contours according to vertical stroke density and transition analysis.
20. The storage medium as defined in claim 15 , wherein the code, when executed, causes the machine to:
combine neighboring keypoint regions with consistent aspect ratios and size to generate a new keypoint region;
detect and combine edge segments of an image in a keypoint region; and
bin sample points on selected edges by angle and distance with reference to a dominant orientation.