Information processing apparatus
An information processing apparatus acquires a plurality of reference images obtained by image capturing of a target which is to be reproduced, acquires a plurality of transformed images obtained by scaling up or down each of the reference images, and generates learned data to be used to generate a reproduction image displaying the external appearance of the target, by performing machine learning using, as teacher data, a plurality of learning target images including the plurality of transformed images.
1. An information processing apparatus, comprising:
a memory storing computer executable instructions; and
a processor configured to execute the computer executable instructions to control the apparatus to:
acquire a plurality of first images of a first user;
generate a plurality of transformed images of the first user, the generating including:
executing a transformation process of scaling up or down each of the plurality of first images of the first user by a plurality of different magnification factors such that a quantity of the plurality of transformed images of the first user is an integer multiple of the plurality of first images of the first user where the integer corresponds to a number of the plurality of different magnification factors,
executing a similarity determination process of classifying one transformed image as having posture data similar to posture data of another transformed image based upon the posture data of the one transformed image and the posture data of the other transformed image meeting a similarity threshold, the similarity threshold representing that postures are more similar than dissimilar, and
executing an exclusion process of excluding, from the plurality of transformed images, the one transformed image classified as having the posture data similar to the posture data of the other transformed image;
generate learned data based on the plurality of transformed images of the first user that was generated through the transformation process, the similarity determination process, and the exclusion process;
acquire a second image of a second user that is different from the first user;
generate posture data indicating a posture of the second user in the second image; and
generate, using the learned data generated based on the plurality of transformed images of the first user and the posture data indicating the posture of the second user, a reproduction image of the first user with the posture of the second user in the second image.
2. The information processing apparatus according to claim 1 , wherein the learned data is generated based on a plurality of learning target images including the plurality of first images of the first user and the plurality of transformed images of the first user that was generated through the transformation process, the similarity determination process, and the exclusion process.
3. The information processing apparatus according to claim 2 , wherein the processor is configured to execute the computer executable instructions to control the apparatus to execute, on each of the learning target images, a process including:
identifying a posture of a target included in the learning target image,
executing machine learning using the posture data and the learning target image, and
generating the learned data.
4. The information processing apparatus according to claim 3 , wherein the processor is configured to execute the computer executable instructions to control the apparatus to exclude from the machine learning a learning target image that is determined to lack a predetermined part of the target as a result of the identifying the posture.
5. The information processing apparatus according to claim 1 , wherein the processor is further configured to execute the computer executable instructions to control the apparatus to:
after executing the transformation process and before executing the similarity determination process, classify the plurality of transformed images of the first user into a plurality of groups each including transformed images exclusively scaled up or down by a same magnification factor of the plurality of different magnification factors;
independently execute the similarity determination process on each of the groups independently; and
independently execute the exclusion process on each of the groups.
6. The information processing apparatus according to claim 1 , wherein the plurality of different magnification factors sequentially change in accordance with a geometric progression.
7. An information processing method, comprising:
acquiring a plurality of first images of a first user;
generating a plurality of transformed images of the first user, the generating including:
executing a transformation process of scaling up or down each of the plurality of first images of the first user by a plurality of different magnification factors such that a quantity of the plurality of transformed images of the first user is an integer multiple of the plurality of first images of the first user where the integer corresponds to a number of the plurality of different magnification factors,
executing a similarity determination process of classifying one transformed image as having posture data similar to posture data of another transformed image based upon the posture data of the one transformed image and the posture data of the other transformed image meeting a similarity threshold, the similarity threshold representing that postures are more similar than dissimilar, and
executing an exclusion process of excluding, from the plurality of transformed images, the one transformed image classified as having the posture data similar to the posture data of the other transformed image;
generating learned data based on the plurality of transformed images of the first user that was generated through the transformation process, the similarity determination process and the exclusion process;
acquiring a second image of a second user that is different from the first user;
generating posture data indicating a posture of the second user in the second image; and
generating, using the learned data generated based on the plurality of transformed images of the first user and the posture data indicating the posture of the second user, a reproduction image of the first user with the posture of the second user in the second image.
8. A non-transitory computer readable medium storing computer executable instructions which, when executed by a processor of an apparatus, cause the apparatus to perform a process comprising:
acquiring a plurality of first images of a first user;
generating a plurality of transformed images of the first user, the generating including:
executing a transformation process of scaling up or down each of the plurality of first images of the first user by a plurality of different magnification factors such that a quantity of the plurality of transformed images of the first user is an integer multiple of the plurality of first images of the first user where the integer corresponds to a number of the plurality of different magnification factors,
executing a similarity determination process of classifying one transformed image as having posture data similar to posture data of another transformed image based upon the posture data of the one transformed image and the posture data of the other transformed image meeting a similarity threshold, the similarity threshold representing that postures are more similar than dissimilar, and
executing an exclusion process of excluding, from the plurality of transformed images, the one transformed image classified as having the posture data similar to the posture data of the other transformed image;
generating learned data based on the plurality of transformed images of the first user that was generated through the transformation process, the similarity determination process and the exclusion process;
acquiring a second image of a second user that is different from the first user;
generating posture data indicating a posture of the second user in the second image; and
generating, using the learned data generated based on the plurality of transformed images of the first user and the posture data indicating the posture of the second user, a reproduction image of the first user with the posture of the second user in the second image.
9. The information processing apparatus according to claim 1 , wherein the processor is configured to execute the computer executable instructions to control the apparatus to:
acquire a plurality of second images of the second user; and
for each second image of the plurality of second images:
generate posture data indicating a posture of the second user in the corresponding second image; and
generate a reproduction image of the first user with the posture of the second user in the corresponding second image.
10. The information processing apparatus according to claim 9 , wherein
the processor is configured to execute the computer executable instructions to control the apparatus to generate a reproduction video, and
frames of the reproduction video include the reproduction images generated for each of the plurality of second images.
11. The information processing apparatus according to claim 1 , wherein the processor is further configured to execute the computer executable instructions to control the apparatus to output the reproduction image.
12. The information processing apparatus according to claim 1 , wherein the reproduction image is output to a display for display by the display.
13. The information processing method according to claim 7 , wherein the learned data is generated based on a plurality of learning target images including the plurality of first images of the first user and the plurality of transformed images of the first user that was generated through the transformation process, the similarity determination process, and the exclusion process.
14. The information processing method according to claim 13 , further comprising:
identifying a posture of a target included in the learning target image,
executing machine learning using the posture data and the learning target image, and
generating the learned data.
15. The information processing method according to claim 14 , further comprising excluding from the machine learning a learning target image that is determined to lack a predetermined part of the target as a result of the identifying the posture.
16. The information processing method according to claim 7 , further comprising:
after executing the transformation process and before executing the similarity determination process, classifying the plurality of transformed images of the first user into a plurality of groups each including transformed images exclusively scaled up or down by a same magnification factor of the plurality of different magnification factors;
independently executing the similarity determination process on each of the groups independently; and
independently executing the exclusion process on each of the groups.
17. The information processing method according to claim 7 , further comprising:
acquiring a plurality of second images of the second user; and
for each second image of the plurality of second images:
generating posture data indicating a posture of the second user in the corresponding second image; and
generating a reproduction image of the first user with the posture of the second user in the corresponding second image.
18. The information processing method according to claim 7 , further comprising outputting the reproduction image.
19. The information processing apparatus according to claim 4 , wherein the processor is configured to execute the computer executable instructions to control the apparatus to exclude from the machine learning only the learning target image that is one of the plurality of transformed images of the first user.
20. The information processing apparatus according to claim 6 , wherein the geometric progression sets each of the plurality of different magnification factors as a function of a maximum magnification factor.