Messaging system with augmented reality makeup
Systems, methods, and computer readable media for messaging system with augmented reality (AR) makeup are presented. Methods include processing a first image to extract a makeup portion of the first image, the makeup portion representing the makeup from the first image and training a neural network to process images of people to add AR makeup representing the makeup from the first image. The methods may further include receiving, via a messaging application implemented by one or more processors of a user device, input that indicates a selection to add the AR makeup to a second image of a second person. The methods may further include processing the second image with the neural network to add the AR makeup to the second image and causing the second image with the AR makeup to be displayed on a display device of the user device.
1. A method, comprising:
accessing a first image of a first person with makeup;
processing the first image, using at least one processor of a user device, to extract a makeup portion of the first image, the makeup portion representing the makeup from the first image;
training a neural network to process images of people to add augmented reality (AR) makeup representing the makeup from the first image, wherein the training is based on comparing the images of people with the AR makeup with the makeup portion of the first image;
receiving, via a messaging application implemented by the at least one processor of the user device, input that indicates a selection to add the AR makeup to a second image of a second person;
processing the second image with the neural network to add the AR makeup to the second image; and
causing the second image with the AR makeup to be displayed on a display of the user device.
2. The method of claim 1 , wherein the second person is a user of the user device.
3. The method of claim 1 , wherein the neural network is a first neural network, and further comprising:
training a second neural network to segment the first image into the makeup portion and a not makeup portion, wherein the second neural network is trained using training data of pairs of images of a person with makeup and a same person without makeup; and wherein the second neural network is used to extract the makeup portion of the first image.
4. The method of claim 3 , wherein the neural network comprises a plurality of maximum pooling operations and then a plurality of up-convolution operations and copy operations, wherein a copy operation of the copy operations appends results from a previous layer of the neural network with the results of an up-convolution operation.
5. The method of claim 4 , wherein the neural network further comprises a fully-connected last layer that indicates segments of the first image as being the makeup portion or the not makeup portion.
6. The method of claim 1 , wherein the training the neural network further comprises:
training the neural network to process images of the people to add AR makeup representing the makeup from the first image, wherein the training is further based on comparing the images of people with the AR makeup to images of people without the AR makeup.
7. The method of claim 6 further comprising:
adjusting weights of the neural network based on backpropagation.
8. The method of claim 1 , wherein the training the neural network further comprises:
determining differences in style between the images of people with the AR makeup and the makeup portion of the first image, wherein the style comprises colors, textures, and common patterns.
9. The method of claim 1 further comprising:
generating an application to process images with the neural network to add the AR makeup to the images; and
storing the application in a database.
10. The method of claim 9 further comprising:
causing to be displayed on the display a list of applications that add AR makeup to images; and
in response to a selection of an application of the applications, processing a live image of the second person with the application to add corresponding AR makeup to the live image.
11. The method of claim 1 further comprising:
causing to be displayed on the display images of people with makeup; the images comprising the first image; and
receiving a selection of the first image; and wherein processing the first image further comprises:
in response to receiving the selection of the first image, processing the first image, using at least one processor, to extract the makeup portion of the first image, the makeup portion representing the makeup from the first image.
12. The method of claim 1 further comprising:
storing in a database an association between weights of the neural network and the first image.
13. The method of claim 1 , wherein the user device is a mobile phone and the second image is a live image of a user of the mobile phone.
14. The method of claim 1 , wherein the neural network is a first neural network, and wherein the method further comprises:
determining a body part of the makeup portion of the first image; and
selecting a second neural network trained to extract makeup portions from the body part, wherein the processing comprises processing using the second neural network.
15. A system comprising:
a processor; and
a memory including instructions that, when executed by the processor, cause the processor to perform operations comprising:
accessing a first image of a first person with makeup;
processing the first image, using at least one processor, to extract a makeup portion of the first image, the makeup portion representing the makeup from the first image; and
training a neural network to process images of people to add augmented reality (AR) makeup representing the makeup from the first image, wherein the training is based on comparing the images of people with the AR makeup with the makeup portion of the first image.
16. The system of claim 15 , wherein the neural network is a first neural network and wherein the instructions further cause the processor to perform operations comprising:
training a second neural network to segment the first image into the makeup portion and a not makeup portion, wherein the second neural network is trained using training data of pairs of images of a person with makeup and a same person without makeup; and wherein the second neural network is used to extract the makeup portion of the first image.
17. The system of claim 15 , wherein the neural network comprises a plurality of maximum pooling operations and then a plurality of up-convolution operations and copy operations, wherein a copy operation of the copy operations appends results from a previous layer of the neural network with the results of an up-convolution operation.
18. The system of claim 15 , wherein the instructions further cause the processor to perform operations comprising:
providing the neural network to a user device.
19. A non-transitory computer-readable medium comprising instructions, which when executed by a computing device comprising one or more processors, cause the one or more processors to:
access a first image of a first person with makeup;
receive, via a messaging application, input that indicates a selection to add augmented reality (AR) makeup to a second image of a second person, the AR makeup representing the makeup from the first image;
access a neural network trained to add the AR makeup representing the makeup from the first image to another image;
process the second image with the neural network to add the AR makeup to the second image; and
causing the second image with the AR makeup to be displayed on a display of the user device.
20. The non-transitory computer-readable medium of claim 19 , wherein the neural network is trained based on comparing images of people with the AR makeup with the makeup portion of the first image.