Font identification from imagery
A system includes a computing device that includes a memory configured to store instructions. The system also includes a processor to execute the instructions to perform operations that include receiving an image that includes textual content in at least one font. Operations also include identifying the at least one font represented in the received image using a machine learning system. The machine learning system being trained using images representing a plurality of training fonts. A portion of the training images includes text located in the foreground and being positioned over captured background imagery.
1. A computing device implemented method comprising:
receiving an image that includes textual content in at least one font; and
identifying the at least one font represented in the received image using a machine learning system, the machine learning system being trained using images representing a plurality of training fonts, wherein a portion of the training images includes synthetic text located in the foreground and being positioned over captured background imagery, and a portion of the training images is distorted when captured by at least one of image capture conditions and capture equipment, wherein the identified at least one font is represented by one element of a plurality of elements of a data vector provided by the machine learning system.
2. The computing device implemented method of claim 1 , wherein the text located in the foreground is synthetically augmented.
3. The computing device implemented method of claim 2 , wherein synthetic augmentation is provided in a two-step process.
4. The computing device implemented method of claim 2 , wherein the text is synthetically augmented based upon one or more predefined conditions.
5. The computing device implemented method of claim 1 , wherein the text located in the foreground is undistorted.
6. The computing device implemented method of claim 1 , wherein the captured background imagery is predominately absent text.
7. The computing device implemented method of claim 1 , wherein the text located in the foreground is randomly positioned in the portion of training images.
8. The computing device implemented method of claim 1 , wherein prior to the text being located in the foreground, a portion of the text is removed.
9. The computing device implemented method of claim 1 , wherein the captured background imagery is distorted when captured.
10. The computing device implemented method of claim 1 , wherein font similarity is used to identify the at least one font.
11. The computing device implemented method of claim 1 , wherein similarity of fonts in multiple image segments is used to identify the at least one font.
12. The computing device implemented method of claim 1 , wherein the machine learning system is trained by using transfer learning.
13. The computing device implemented method of claim 1 , wherein an output of the machine learning system represents each font used to train the machine learning system.
14. The computing device implemented method of claim 13 , wherein the output of the machine learning system provides a level of confidence for each font used to train the machine learning system.
15. The computing device implemented method of claim 1 , wherein a subset of the output of the machine learning system is scaled and a remainder of the output is removed.
16. The computing device implemented method of claim 1 , wherein some of the training images are absent identification.
17. The computing device implemented method of claim 1 , wherein identifying the at least one font represented in the received image using the machine learning system includes using additional images received by the machine learning system.
18. The computing device implemented method of claim 17 , wherein outputs of the machine learning system for the received image and the additional images are combined to identify the at least one font.
19. The computing device implemented method of claim 1 , wherein the machine learning system comprises a generative adversarial network (GAN).
20. The computing device implemented method of claim 19 , wherein the generative adversarial network (GAN) comprises a generator neural network and a discriminator neural network.
21. A system comprising:
a computing device comprising:
a memory configured to store instructions; and
a processor to execute the instructions to perform operations comprising:
receiving an image that includes textual content in at least one font; and
identifying the at least one font represented in the received image using a machine learning system, the machine learning system being trained using images representing a plurality of training fonts, wherein a portion of the training images includes synthetic text located in the foreground and being positioned over captured background imagery, and a portion of the training images is distorted when captured by at least one of image capture conditions and capture equipment, wherein the identified at least one font is represented by one element of a plurality of elements of a data vector provided by the machine learning system.
22. The system of claim 21 , wherein the text located in the foreground is synthetically augmented.
23. The system of claim 22 , wherein synthetic augmentation is provided in a two-step process.
24. The computing device implemented method of claim 22 , wherein the text is synthetically augmented based upon one or more predefined conditions.
25. The system of claim 21 , wherein the text located in the foreground is undistorted.
26. The system of claim 21 , wherein the captured background imagery is predominately absent text.
27. The system of claim 21 , wherein the text located in the foreground is randomly positioned in the portion of training images.
28. The system of claim 21 , wherein prior to the text being located in the foreground, a portion of the text is removed.
29. The system of claim 21 , wherein the captured background imagery is distorted when captured.
30. The system of claim 21 , wherein font similarity is used to identify the at least one font.
31. The system of claim 21 , wherein similarity of fonts in multiple image segments is used to identify the at least one font.
32. The system of claim 21 , wherein the machine learning system is trained by using transfer learning.
33. The system of claim 21 , wherein an output of the machine learning system represents each font used to train the machine learning system.
34. The system of claim 33 , wherein the output of the machine learning system provides a level of confidence for each font used to train the machine learning system.
35. The system of claim 21 , wherein a subset of the output of the machine learning system is scaled and a remainder of the output is removed.
36. The system of claim 21 , wherein some of the training images are absent identification.
37. The system of claim 21 , wherein identifying the at least one font represented in the received image using the machine learning system includes using additional images received by the machine learning system.
38. The system of claim 37 , wherein outputs of the machine learning system for the received image and the additional images are combined to identify the at least one font.
39. The system of claim 21 , wherein the machine learning system comprises a generative adversarial network (GAN).
40. The system of claim 39 , wherein the generative adversarial network (GAN) comprises a generator neural network and a discriminator neural network.
41. One or more non-transitory computer readable media storing instructions that are executable by a processing device, and upon such execution cause the processing device to perform operations comprising:
receiving an image that includes textual content in at least one font; and
identifying the at least one font represented in the received image using a machine learning system, the machine learning system being trained using images representing a plurality of training fonts, wherein a portion of the training images includes synthetic text located in the foreground and being positioned over captured background imagery, and a portion of the training images is distorted when captured by at least one of image capture conditions and capture equipment, wherein the identified at least one font is represented by one element of a plurality of elements of a data vector provided by the machine learning system.
42. The non-transitory computer readable media of claim 41 , wherein the text located in the foreground is synthetically augmented.
43. The non-transitory computer readable media of claim 42 , wherein synthetic augmentation is provided in a two-step process.
44. The non-transitory computer readable media of claim 42 , wherein the text is synthetically augmented based upon one or more predefined conditions.
45. The non-transitory computer readable media of claim 41 , wherein the text located in the foreground is undistorted.
46. The non-transitory computer readable media of claim 41 , wherein the captured background imagery is predominately absent text.
47. The non-transitory computer readable media of claim 41 , wherein the text located in the foreground is randomly positioned in the portion of training images.
48. The non-transitory computer readable media of claim 41 , wherein prior to the text being located in the foreground, a portion of the text is removed.
49. The non-transitory computer readable media of claim 41 , wherein the captured background imagery is distorted when captured.
50. The non-transitory computer readable media of claim 41 , wherein font similarity is used to identify the at least one font.
51. The non-transitory computer readable media of claim 41 , wherein similarity of fonts in multiple image segments is used to identify the at least one font.
52. The non-transitory computer readable media of claim 41 , wherein the machine learning system is trained by using transfer learning.
53. The non-transitory computer readable media of claim 41 , wherein an output of the machine learning system represents each font used to train the machine learning system.
54. The non-transitory computer readable media of claim 53 , wherein the output of the machine learning system provides a level of confidence for each font used to train the machine learning system.
55. The non-transitory computer readable media of claim 41 , wherein a subset of the output of the machine learning system is scaled and a remainder of the output is removed.
56. The non-transitory computer readable media of claim 41 , wherein some of the training images are absent identification.
57. The non-transitory computer readable media of claim 41 , wherein identifying the at least one font represented in the received image using the machine learning system includes using additional images received by the machine learning system.
58. The non-transitory computer readable media of claim 41 , wherein outputs of the machine learning system for the received image and the additional images are combined to identify the at least one font.
59. The non-transitory computer readable media of claim 41 , wherein the machine learning system comprises a generative adversarial network (GAN).
60. The non-transitory computer readable media of claim 59 , wherein the generative adversarial network (GAN) comprises a generator neural network and a discriminator neural network.