Abstract: Gauging consumer interest of in-person visitors is described. In one or more embodiments, a promotional apparatus having a display screen and a wireless beacon is deployed at a venue. The promotional apparatus presents a video having multiple video segments. During the presentation, the beacon and a mobile device of a venue visitor communicate wirelessly to determine a proximity of the visitor to the display screen. Multiple distances are determined at multiple times. A path of movement can be determined from the distances over time. If the visitor alters a direction of travel to veer toward the screen or stops walking close to the screen, a system infers that the video has engaged the visitor's attention. By linking particular video segments to times at which the mobile device is proximate to the display screen, the system can further infer consumer interest in specific product features being shown at those times.

Abstract: Systems and methods for removing objects from images are disclosed. An image processing application identifies a boundary of each object of a set of objects in an image. In some cases, the identification uses deep learning. The image processing application identifies a completed boundary for each object of the set of objects by providing the object to a trained model. The image processing application determines a set of masks. Each mask corresponds to an object of the set of objects and represents a region of the image defined by an intersection of the boundary of the object and the boundary of a target object to be removed from the image. The image processing application updates each mask by separately performing content filling on the corresponding region. The image processing application creates an output image by merging each of the updated masks with portions of the image.

Abstract: Systems and techniques for masking non-public content in screen images are provided. An example system includes a screen capture tool, a region-based object detection system, a classifier, and an image masking engine. The screen capture tool may be configured to generate a screen image representing a screen being displayed by the system. The region-based object detection system may be configured to identify multiple regions within the screen image as potential non-public content regions. The classifier may be configured to selectively classify the identified regions as non-public content regions. The image masking engine may be configured to generate a masked image by masking the regions classified as non-public content regions in the screen image.

Abstract: Methods, systems, and software are disclosed herein for protecting the privacy of users in images and to prevent or mitigate the occurrence of privacy breaches. In an implementation, a biometric privacy service identifies a biometric marker in an image of a scene, such as a fingerprint, a retina, or an iris. The service then identifies one or more features of the biometric marker to be modified. The biometric service also identifies one or more modifications to make to the one or more features of the biometric marker and generates a modified version of the biometric marker that reflects the modification(s). The image is updated by the service with the modified version of the biometric marker, so that the biometric marker can no longer be used for unwanted purposes, whether they be malicious in intent or benign.

Abstract: An interactive email experience is customized to the recipient's interests by modifying rich media components provided by the email based on the recipient's interactions with other rich media components from the email. To facilitate the interactive email experience, rich media components are provided by a marketer for an email campaign with mapping information mapping product features to portions of the rich media components. When an email is sent with links to the rich media components, the recipient's interactions with a rich media component is tracked. Product features are ranked based on the recipient's interactions with various portions corresponding with the various product features. The product feature rankings are then used to modify other rich media components from the email to emphasize portions of the other rich media components corresponding with product features of interest to the recipient.

Abstract: Background content suggestion for combination with identified items is described. Initially, a system receives digital visual content depicting an item, e.g., a product for purchase. The system identifies various content items that are available for suggesting as replacement backgrounds for the item and then determines a compatibility of these identified content items with the item depicted in the received digital visual content. In particular, the system determines compatibility based on both a scene compatibility and a color compatibility of the identified content items with the depicted item. Based on a combination of the scene and color compatibility, the system surfaces at least one of the identified content items (e.g., a highest scoring content item) as a suggested replacement background for the depicted item.

Abstract: The present disclosure relates to systems, methods, and non-transitory computer readable media for generating target products for a product search based on gesture input received via a digital canvas. For example, the disclosed systems can utilize digital image classification models to generate product sets based on individual visual product features of digital images of products. The disclosed systems can further receive gesture input within a digital canvas indicating visual product features. In addition, the disclosed systems can compare the gesture input of the digital canvas with representative digital images of product sets generated by particular classification models to identify product sets that include the indicated visual product features. Further, the disclosed systems can provide target products from the identified product sets for display via a product search interface website.

Abstract: Systems and methods are provided for intelligently adapting content presented in electronic forms. In some embodiments, a processor provides access to multiple instances of a first version of an electronic form by multiple clients based on determining that insufficient data exists in a data source for a category of information. The first version of the form can include content for soliciting data associated with the category of information. The processor can receive responsive electronic communications in response to providing access to the first version of the form. The processor can provide access to multiple instances of a second version of the electronic form by multiple clients based on determining from the received responses whether sufficient data exists for the category of information. The second version of the electronic form can omit or replace the content for soliciting data associated with the category of information.

Abstract: A media player may receive an instruction to advance the media file content a higher than normal speed. At least a portion of the associated metadata may be accessed to retrieve a location of a group of words determined to be the most important in a portion of the media file content being advanced at the higher than normal speed. Audio content of the group of the most important words may be played at a normal speed while advancing the media file at a higher than normal speed.

Abstract: Navigating a video recording based on changes in views of the recording's visual content is described. A content-based navigation system receives a recording including visual content and audio content. The content-based navigation system then determines a content scale for navigating the recording relative to an overall number of new or updated views of visual content during playback of the recording. Given the content scale, the content-based navigation system generates a content navigation control that enables navigating the recording at a granularity defined by the overall number of new or updated views of the recording's visual content. Navigation via the content navigation control is thus independent of time between changes to views of the recording's visual content during playback. Input to the content navigation control causes output of a different view of the recording's visual content, and optionally causes output of audio content synchronized with the different view of visual content.

Abstract: In implementations of a visitor engagement detection and control system for online sessions with a listing platform, a computing device implements a visitor engagement system to monitor user actions by the visitors with items listed by the listing platform, and map the user actions and the items listed by the listing platform to a timeline of the online session. An engagement module determines levels of visitor engagement in the online session based on the mapping of the user actions and the items, and initiates communication with a low-engagement visitor during the online session to increase an engagement level of the low-engagement visitor. In some cases, a communication is then established between the low-engagement visitor and a different low-engagement visitor, an engaged visitor, or an administrator of the listing platform.

Abstract: Methods select content to be delivered to a visitor to a real-world venue. One method identifies a visitor to a real-world venue based on tracked physical actions and a profile including previously collected visitor information associated with the venue. The method determines that members of a group are present at the venue with the visitor, where the determining is based at least in part on the previously collected visitor information. Based on determining that the visitor is at the venue with members of the group, content is selected for the visitor. Another method identifies a visitor to a venue as belonging to a group by identifying the visitor based on tracked physical actions and collected visitor information. The tracked actions and visitor information are used to determine whether the visitor is in the venue with members of the group. Tailored content is sent to the visitor based on the determination.

Abstract: The present disclosure is directed towards methods and systems for providing relevant video scenes in response to a video search query. The systems and methods identify a plurality of key frames of a media object and detect one or more content features represented in the plurality of key frames. Based on the one or more detect content features, the systems and methods associate tags indicating the detected content features with the plurality of key frames of the media object. The systems and methods, in response to receiving a search query including search terms, compare the search terms with the tags of the selected key frames, identify a selected key frame that depicts at least one content feature related to the search terms, and provide a preview image of the media item depicting the at least one content feature.

Abstract: Systems and methods for removing objects from images are disclosed. An image processing application identifies a boundary of each object of a set of objects in an image. In some cases, the identification uses deep learning. The image processing application identifies a completed boundary for each object of the set of objects by providing the object to a trained model. The image processing application determines a set of masks. Each mask corresponds to an object of the set of objects and represents a region of the image defined by an intersection of the boundary of the object and the boundary of a target object to be removed from the image. The image processing application updates each mask by separately performing content filling on the corresponding region. The image processing application creates an output image by merging each of the updated masks with portions of the image.

Abstract: In implementations of jointly editing related objects in a digital image, an image editing application segments a first object in a digital image, and an additional object corresponding to the first object, such a shadow cast by the first object, a reflection of the first object, or an object of a same object class as the first object. Respective stick diagrams for the first object and the additional object are generated, and a mapping of the first object to the additional object is generated based on the stick diagrams. Based on a user request to edit the first object, such as to warp the first object, the first object and the additional object are jointly edited based on the mapping. Accordingly, realistic digital images are efficiently generated that maintain correspondence between related objects, without requiring a user to edit each object individually, thus saving time and resources.

Abstract: Offline consumer behavior and interactions are observed using beacons positioned at venues where consumers spend time and interact with each other. The beacons communicate with mobile computing devices that are carried by the observed consumers. The observed consumer behavior provides actionable insights into how consumers influence each other. For example, the people that a particular consumer spends time with form a “circle of influence” associated with that consumer. If the consumer makes a purchase, members of the circle of influence are observed to understand the degree to which they were influenced by the purchase, if at all. Metrics that quantify a consumer's influence over other consumers allow marketers to more effectively target both the influencing and influenced consumers. Also, if relatively little information is known about a particular consumer, that consumer's digital marketing profile can be supplemented using information known about the consumers with whom he/she often spends time.

Abstract: A digital medium environment is described to recommend a transition from use of a limited-functionality application to a full-functionality application. As described herein, the limited-functionality application includes a subset of the functionalities of the full-functionality application. Interactions by a user with a limited-functionality application to generate or edit a creation (e.g., a photo, video, or document) are monitored. Based on the interactions, it is determined that the user needs a functionality for the creation that is not available in the limited-functionality application but is supported by the full-functionality application. A recommendation to transition the creation to the full-functionality application that supports the functionality is displayed. The recommendation may include a selectable control that, when selected, causes the creation to be transitioned to the full-functionality application to enable the functionality to be applied to the creation.

Abstract: In implementations of a participant engagement detection and control system for online sessions, a computing device implements a participant engagement system to monitor user actions of participants during an online session, and track engagement levels of the participants by mapping the user actions and content of the online session to a timeline of the online session. Engagement scores for the participants are determined based on the mapping of the user actions and the content at each time interval of the timeline. Low-engagement participants having a probability of dropping out of the online session are determined based on the engagement scores. Engaging content is selected as determined to increase an engagement level of the low-engagement participants, and the engaging content is communicated to the low-engagement participants. Furthermore, the engaging content and identifiers of the low-engagement participants can be communicated to a presenter of the online session.

Abstract: Computer-implemented systems and methods disclosed herein disclose features for determining the frequency of exposure to real-world products embedded with Internet-Of-Things (IoT) hardware and determining a comprehensive targeting profile identifying the interests of users based on their exposure to real-world products. A computing device carried by a user wirelessly receives product exposure data from a nearby IoT-enabled device upon entering within a certain proximity of the IoT-enabled device. The computing device further compiles the product exposure data with previous product exposure data for IoT-enabled devices to which the computing device was previously exposed. Based on the compiled product exposure data, a server device determines a frequency of exposure to a product corresponding to the IoT-enabled device over a period of time.

Abstract: Certain embodiments involve optimizing how interactive content is arranged in electronic communications. A processor identifies a subject matter type associated with first and second online content items having respective first and second presentation types. The processor determines that a client has previously interacted more extensively with online content having the subject matter type and the first presentation type as compared to other online content having the subject matter type and the second presentation type. The processor selects a layout for an electronic communication for accessing the online content items based on determining that the client interacted more extensively with online content having the first presentation type. The layout displays a first portion for accessing the first online content item more prominently than a second portion for accessing the second online content item. The processor presents the layout via an interface for editing the electronic communication.