Machine learning-based modification of image content
Aspects of the present disclosure involve a system comprising a computer-readable storage medium storing a program and method for modifying a captured image. The program and method provide for displaying, by a messaging application, an image captured by a device camera; providing, by the messaging application, a user interface for selecting from among a plurality of content modifiers to modify the image, the plurality of content modifiers including a first content modifier corresponding to a machine learning model trained with a plurality of image pairs, each image pair including a first image and a second image corresponding to a modified version of the first image; receiving user selection of the first content modifier from among the plurality of content modifiers; determining, in response to receiving the user selection, a modified version of the image based on output from the machine learning model; and displaying the modified version of the image.
1 . A method, comprising:
providing, by a messaging application running on a device, a message preview interface for displaying an image captured by a camera of the device,
wherein the message preview interface includes a set of editing tools, each tool within the set of editing tools being represented by a respective icon which is user selectable to annotate or modify the image, and
wherein the message preview interface further includes a plurality of content modifiers is separate from the set of editing tools, each content modifier within the plurality of content modifiers corresponding to respective augmented reality content that is automatically applied to the image via respective swipe gestures to individually cycle through different augmented reality content to apply to the image, the respective swipe gestures being performed directly on the image, and
wherein the plurality of content modifiers includes a first content modifier configured to apply augmented reality content to the image using a machine learning model, the machine learning model having been trained with a plurality of image pairs, each image pair including a first image and a second image corresponding to a modified version of the first image;
receiving, by the messaging application, a swipe gesture selecting the first content modifier from among the plurality of content modifiers;
determining, in response to receiving the swipe gesture, a modified version of the image based on output from the machine learning model; and
displaying the modified version of the image.
2 . The method of claim 1 , wherein the machine learning model receives the image as input and provides the modified version of the image as the output.
3 . The method of claim 1 , wherein determining the modified version of the image comprises:
sending, to a server which implements the first content modifier, the image and a request for the modified version of the image; and
receiving, from the server and in response to the request, the modified version of the image.
4 . The method of claim 1 , wherein the first content modifier corresponds to a filter for providing an overlay on the image.
5 . The method of claim 1 , wherein the first content modifier corresponds to an augmented reality content item for displaying augmented reality content with the image.
6 . The method of claim 1 , wherein the message preview interface provides for navigating through the plurality of content modifiers and displaying a manner in which each respective content modifier modifies the image.
7 . The method of claim 1 , wherein the plurality of image pairs are based on a predetermined set of image pairs selected using human evaluation.
8 . The method of claim 7 , wherein the plurality of image pairs are further based on user-submitted feedback provided with respect to the predetermined set of image pairs, the user-submitted feedback corresponding to at least one of crowd-sourced feedback or feedback provided by a user of the device.
9 . A device, comprising:
a processor; and
a memory storing instructions that, when executed by the processor, cause the processor to:
provide, by a messaging application running on the device, a message preview interface for displaying an image captured by a camera of the device,
wherein the message preview interface includes a set of editing tools, each tool within the set of editing tools being represented by a respective icon which is user selectable to annotate or modify the image, and
wherein the message preview interface further includes a plurality of content modifiers is separate from the set of editing tools, each content modifier within the plurality of content modifiers corresponding to respective augmented reality content that is automatically applied to the image via respective swipe gestures to individually cycle through different augmented reality content to apply to the image, the respective swipe gestures being performed directly on the image, and
wherein the plurality of content modifiers includes a first content modifier configured to apply augmented reality content to the image using a machine learning model, the machine learning model having been trained with a plurality of image pairs, each image pair including a first image and a second image corresponding to a modified version of the first image;
receiving, by the messaging application, a swipe gesture selecting the first content modifier from among the plurality of content modifiers;
determining, in response to receiving the swipe gesture, a modified version of the image based on output from the machine learning model; and
displaying the modified version of the image.
10 . The device of claim 9 , wherein the machine learn model receives the image as input and provides the modified version of the image as the output.
11 . The device of claim 9 , wherein determining the modified version of the image comprises:
sending, to a server which implements the first content modifier, the image and a request for the modified version of the image; and
receiving, from the server and in response to the request, the modified version of the image.
12 . The device of claim 9 , wherein the first content modifier corresponds to a filter for providing an overlay on the image.
13 . The device of claim 9 , wherein the first content modifier corresponds to an augmented reality content item for displaying augmented reality content with the image.
14 . The device of claim 9 , wherein the message preview interface provides for navigating through the plurality of content modifiers and display a manner in which each respective content modifier modifies the image.
15 . The device of claim 9 , wherein the plurality of image pairs are based on a predetermined set of image pairs selected use human evaluation.
16 . The device of claim 15 , wherein the plurality of image pairs are further based on user-submitted feedback provided with respect to the predetermined set of image pairs, the user-submitted feedback corresponding to at least one of crowd-sourced feedback or feedback provided by a user of the device.
17 . A non-transitory computer-readable storage medium, the computer-readable storage medium including instructions that when executed by a computer, cause the computer to:
provide, by a messaging application running on a device, a message preview interface for displaying an image captured by a camera of the device,
wherein the message preview interface includes a set of editing tools, each tool within the set of editing tools being represented by a respective icon which is user selectable to annotate or modify the image, and
wherein the message preview interface further includes a plurality of content modifiers is separate from the set of editing tools, each content modifier within the plurality of content modifiers corresponding to respective augmented reality content that is automatically applied to the image via respective swipe gestures to individually cycle through different augmented reality content to apply to the image, the respective swipe gestures being performed directly on the image, and
wherein the plurality of content modifiers includes a first content modifier configured to apply augmented reality content to the image using a machine learning model, the machine learning model having been trained with a plurality of image pairs, each image pair including a first image and a second image corresponding to a modified version of the first image;
receiving, by the messaging application, a swipe gesture selecting the first content modifier from among the plurality of content modifiers;
determining, in response to receiving the swipe gesture, a modified version of the image based on output from the machine learning model; and
displaying the modified version of the image.
18 . The computer-readable storage medium of claim 17 , wherein the machine learn model receives the image as input and provides the modified version of the image as the output.
19 . The method of claim 1 , wherein each content modifier within the plurality of content modifiers is not represented by a respective icon.
20 . The method of claim 1 , wherein the plurality of content modifiers is separate from the set of editing tools such that applying any of the plurality of content modifiers to the image does not require selection of an editing tool from the set of editing tools.