Abstract: Among other things, embodiments of the present disclosure improve the functionality of electronic messaging and imaging software and systems by enabling users to generate customized media overlays that can be shared with other users. For example, media overlays can be generated by the system and displayed in conjunction with media content (e.g., images and/or video) generated by an image-capturing device (e.g., a digital camera). In some embodiments, existing media overlays may be used by users to create derivative media overlays. The system may track usage of media overlays and any derivatives created based thereon, and allow users to control the distribution and use of their overlays in future derivatives. In some embodiments, for example, a user can modify an overlay they created and cause the modification to propagate to all derivative overlays based on the user's overlay.

Abstract: Disclosed are systems, methods, and computer-readable storage media to perform an interactive image recolorization process. The method includes receiving user input including a single stroke drawn on an image presented on a client device. The method further includes expanding the single stroke drawn on the image to generate an expanded stroke mask and refining the expanded stroke mask to determine a precise boundary that defines a region of interest on the image. The method further includes recolorizing the region of interest on the image in accordance with a user-specified color and causing presentation of a result of the recolorization on the client device.

Abstract: Disclosed are systems, methods, and computer-readable storage media to perform automatic image inpainting using local patch statistics. The method includes receiving a user input identifying a user-selected region of an image and determining a local region for the user-selected region. The local region includes a portion of the image outside of the user-selected region and that surrounds the user-selected region. The method may further include enlarging the local region by a predefined padding size, and scaling the user-selected region to a predetermined size. The method may further include identifying patch matches within the local region, and inpainting the user-selected region using some of the patch matches identified within the local region.

Abstract: Among other things, embodiments of the present disclosure improve the functionality of electronic messaging and imaging software and systems by determining topics of electronic communications between users and generating customized media content items based on such topics. The media content can be generated for a variety of topics and shared with other users. For example, media content (e.g., images or video can be generated and displayed on a user's computing device, as well as transmitted to other users via electronic communications, such as short message service (SMS) or multimedia service (MMS) texts and emails.

Abstract: Among other things, embodiments of the present disclosure improve the functionality of electronic messaging and imaging software and systems by enabling users to generate customized media overlays that can he shared with other users. For example, media overlays can be generated by the system and displayed in conjunction with media content (e.g., images and/or video) generated by an image-capturing device (e.g., a digital camera). In some embodiments, existing media overlays may be used by users to create derivative media overlays. The system may track usage of media overlays and any derivatives created based thereon, and allow users to control the distribution and use of their overlays in future derivatives. In some embodiments, for example, a user can modify an overlay they created and cause the modification to propagate to all derivative overlays based on the user's overlay.

Abstract: Systems, methods, and non-transitory computer-readable media can obtain a first batch of content items to be clustered. A set of clusters can be generated by clustering respective binary hash codes for each content item in the first batch, wherein content items included in a cluster are visually similar to one another. A next batch of content items to be clustered can be obtained. One or more respective binary hash codes for the content items in the next batch can be assigned to a cluster in the set of clusters.

Abstract: Systems, methods, and non-transitory computer-readable media can receive a compressed convolutional neural network (CNN). A media content item to be processed can be acquired. The compressed CNN to can be utilized to apply a media processing technique to the media content item to produce information about the media content item. It can be determined, based on at least some of the information about the media content item, whether to transmit at least a portion of the media content item to one or more remote servers for additional media processing.

Abstract: Systems and techniques are provided for a ranking approach to train deep neural nets for multilabel image annotation. Label scores may be received for labels determined by a neural network for training examples. Each label may be a positive label or a negative label for the training example. An error of the neural network may be determined based on a comparison, for each of the training examples, of the label scores for positive labels and negative labels for the training example and a semantic distance between each positive label and each negative label for the training example. Updated weights may be determined for the neural network based on a gradient of the determined error of the neural network. The updated weights may be applied to the neural network to train the neural network.

Abstract: Separately optimizing angle error and magnitude error of a search query entered into a query database may be referred to as the “shape-gain” separation quantization. Each of a direction and a magnitude for each of a plurality of database vectors may be separately encoded. A query vector may be received. The query vector may include a query direction and a query magnitude. The separately encoded query direction, query magnitude, and each of the separately encoded direction and magnitude for each of the plurality of database vectors may be combined. Distances between the query vector and each of the plurality of database vectors may be determined. At least one of the plurality of database vectors that is similar to the query vector may be identified based on the determined distances.