Machine learning based image processing techniques
A machine learning based image processing architecture and associated applications are disclosed herein. In some embodiments, a machine learning framework is trained to learn low level image attributes such as object/scene types, geometries, placements, materials and textures, camera characteristics, lighting characteristics, contrast, noise statistics, etc. Thereafter, the machine learning framework may be employed to detect such attributes in other images and process the images at the attribute level.
1. A method, comprising:
receiving a set of images having a prescribed artist imparted aesthetic;
identifying a set of attributes comprising the artist imparted aesthetic using a machine learning framework trained on image datasets comprising a constrained set of objects associated with a prescribed scene type to which the set of images belongs; and
generating one or more additional images having the artist imparted aesthetic using the identified set of attributes and without artist input.
2. The method of claim 1 , wherein the artist imparted aesthetic is associated with the prescribed scene type.
3. The method of claim 1 , wherein the artist imparted aesthetic comprises a distinct signature style.
4. The method of claim 1 , wherein the artist imparted aesthetic comprises a specific branded visual appearance.
5. The method of claim 1 , wherein the artist imparted aesthetic results from artist post-processing to manually create imagery having visual characteristics that conform to a desired or sanctioned style or theme.
6. The method of claim 1 , wherein the artist imparted aesthetic results from artist manipulation beyond what can be achieved from photography, rendering, or global post-processing.
7. The method of claim 1 , wherein the set of images comprises curated images.
8. The method of claim 1 , wherein images of the set of images are not labeled or tagged.
9. The method of claim 1 , wherein images of the set of images comprise different subsets of the set of objects but share the same high level style or visual appearance.
10. The method of claim 1 , wherein images of the set of images are visually recognizable as belonging to the same set.
11. The method of claim 1 , wherein the artist imparted aesthetic comprises a nonlinear function of the set of attributes.
12. The method of claim 1 , wherein the image datasets on which the machine learning framework is trained are labeled or tagged with metadata.
13. The method of claim 1 , further comprising labeling or tagging images of the set of images with the identified set of attributes.
14. The method of claim 1 , wherein the one or more additional images are generated by applying the identified set of attributes to three-dimensional object models that are used to render the additional images.
15. The method of claim 1 , wherein the set of attributes comprises one or more attributes associated with object/scene types, geometries, placements, materials, textures, camera characteristics, lighting characteristics, noise statistics, and contrast.
16. The method of claim 1 , wherein images of the set of images and the generated one or more additional images comprise photographs or photorealistic renderings.
17. The method of claim 1 , wherein images of the set of images and the generated one or more additional images comprise frames of an animation or a video sequence.
18. A system, comprising:
a processor configured to:
receive a set of images having a prescribed artist imparted aesthetic;
identify a set of attributes comprising the artist imparted aesthetic using a machine learning framework trained on image datasets comprising a constrained set of objects associated with a prescribed scene type to which the set of images belongs; and
generate one or more additional images having the artist imparted aesthetic using the identified set of attributes and without artist input; and
a memory coupled to the processor and configured to provide the processor with instructions.
19. The system of claim 18 , wherein the artist imparted aesthetic is associated with the prescribed scene type.
20. The system of claim 18 , wherein the artist imparted aesthetic comprises a distinct signature style.
21. The system of claim 18 , wherein the artist imparted aesthetic comprises a specific branded visual appearance.
22. The system of claim 18 , wherein the artist imparted aesthetic results from artist post-processing to manually create imagery having visual characteristics that conform to a desired or sanctioned style or theme.
23. The system of claim 18 , wherein the artist imparted aesthetic results from artist manipulation beyond what can be achieved from photography, rendering, or global post-processing.
24. The system of claim 18 , wherein the set of images comprises curated images.
25. The system of claim 18 , wherein images of the set of images are not labeled or tagged.
26. The system of claim 18 , wherein images of the set of images comprise different subsets of the set of objects but share the same high level style or visual appearance.
27. The system of claim 18 , wherein images of the set of images are visually recognizable as belonging to the same set.
28. The system of claim 18 , wherein the artist imparted aesthetic comprises a nonlinear function of the set of attributes.
29. The system of claim 18 , wherein the image datasets on which the machine learning framework is trained are labeled or tagged with metadata.
30. The system of claim 18 , wherein the processor is further configured to label or tag images of the set of images with the identified set of attributes.
31. The system of claim 18 , wherein the one or more additional images are generated by applying the identified set of attributes to three-dimensional object models that are used to render the additional images.
32. The system of claim 18 , wherein the set of attributes comprises one or more attributes associated with object/scene types, geometries, placements, materials, textures, camera characteristics, lighting characteristics, noise statistics, and contrast.
33. The system of claim 18 , wherein images of the set of images and the generated one or more additional images comprise photographs or photorealistic renderings.
34. The system of claim 18 , wherein images of the set of images and the generated one or more additional images comprise frames of an animation or a video sequence.
35. A computer program product embodied in a non-transitory computer readable storage medium and comprising computer instructions for:
receiving a set of images having a prescribed artist imparted aesthetic;
identifying a set of attributes comprising the artist imparted aesthetic using a machine learning framework trained on image datasets comprising a constrained set of objects associated with a prescribed scene type to which the set of images belongs; and
generating one or more additional images having the artist imparted aesthetic using the identified set of attributes and without artist input.
36. The computer program product of claim 35 , wherein the artist imparted aesthetic is associated with the prescribed scene type.
37. The computer program product of claim 35 , wherein the artist imparted aesthetic comprises a distinct signature style.
38. The computer program product of claim 35 , wherein the artist imparted aesthetic comprises a specific branded visual appearance.
39. The computer program product of claim 35 , wherein the artist imparted aesthetic results from artist post-processing to manually create imagery having visual characteristics that conform to a desired or sanctioned style or theme.
40. The computer program product of claim 35 , wherein the artist imparted aesthetic results from artist manipulation beyond what can be achieved from photography, rendering, or global post-processing.
41. The computer program product of claim 35 , wherein the set of images comprises curated images.
42. The computer program product of claim 35 , wherein images of the set of images are not labeled or tagged.
43. The computer program product of claim 35 , wherein images of the set of images comprise different subsets of the set of objects but share the same high level style or visual appearance.
44. The computer program product of claim 35 , wherein images of the set of images are visually recognizable as belonging to the same set.
45. The computer program product of claim 35 , wherein the artist imparted aesthetic comprises a nonlinear function of the set of attributes.
46. The computer program product of claim 35 , wherein the image datasets on which the machine learning framework is trained are labeled or tagged with metadata.
47. The computer program product of claim 35 , further comprising computer instructions for labeling or tagging images of the set of images with the identified set of attributes.
48. The computer program product of claim 35 , wherein the one or more additional images are generated by applying the identified set of attributes to three-dimensional object models that are used to render the additional images.
49. The computer program product of claim 35 , wherein the set of attributes comprises one or more attributes associated with object/scene types, geometries, placements, materials, textures, camera characteristics, lighting characteristics, noise statistics, and contrast.
50. The computer program product of claim 35 , wherein images of the set of images and the generated one or more additional images comprise photographs or photorealistic renderings.
51. The computer program product of claim 35 , wherein images of the set of images and the generated one or more additional images comprise frames of an animation or a video sequence.