Abstract: Described herein are systems and methods that highlight target objects in media content. The detection system trains a neural network to identify objects within an image. The detection system receives a text input requesting a search within the image and applies a search language model to the text input, which identifies a target object associated with the requested search. The detection system applies the neural network to the image to identify instances of the target object. The detection system modifies a user interface to include the image and modifies the image to highlight the identified instances of the target object. The detection system receives feedback that modifies the highlighted instances of the target object within the user interface and retrains the neural network based on the modified highlighted instances.

Abstract: A system accesses a first video stream from an internal scanning device (e.g., an X-ray scanner) that scans objects or individuals. It also accesses a second video stream from a capturing device that records a human operator reviewing and interacting with the first stream on a display to identify targeted subject matter. The system then identifies the targeted subject matter based on the operator's interactions and constructs a training dataset based on the identified targeted subject matter. Using this training dataset, the system trains a machine-learning model to identify the targeted subject matter in future video streams from scanning devices.

Abstract: A media detection system receives a video corresponding to a fixed field of view. The media detection system may receive user input indicating one or more object types to identify or a subset of the video within which to identify objects. The media detection system applies one or more machine-learned classifiers to frames of the video and creates a summary video that includes the background of the video and identified instances for simultaneous playback within the fixed field of view. The media detection system may also identify instances of objects in a live video stream and use the identified instances to respond to user questions. The media detection system applies a language model to questions to identify the subject matter of the questions, identifies content within the live video stream associated with the subject matter, and uses the identified content to respond to the user's question.

Abstract: A media detection system receives a video corresponding to a fixed field of view. The media detection system may receive user input indicating one or more object types to identify or a subset of the video within which to identify objects. The media detection system applies one or more machine-learned classifiers to frames of the video and creates a summary video that includes the background of the video and identified instances for simultaneous playback within the fixed field of view. The media detection system may also identify instances of objects in a live video stream and use the identified instances to respond to user questions. The media detection system applies a language model to questions to identify the subject matter of the questions, identifies content within the live video stream associated with the subject matter, and uses the identified content to respond to the user's question.

Abstract: Described herein are systems and methods that detect and filter objects from media content. In particular, a detection system accesses classifiers that can detect objects when applied to a video. The detection system receives a text input from a user interface identifying filtering criteria for outputs of the classifiers and inputs the filtering criteria to a filtering language model that produces a filtering object. The detection system presents the filtering object at the user interface. In response to a user accepting the filtering object, the detection system applies the classifiers to the video, which produces a set of detected objects. The detection system applies the filtering object to the detected objects by removing a subset of the detected objects that do not satisfy the filtering criteria and presents the video with the filtered set of detected objects highlighted within the video.

Abstract: A media detection system receives a video corresponding to a fixed field of view. The media detection system may receive user input indicating one or more object types to identify or a subset of the video within which to identify objects. The media detection system applies one or more machine-learned classifiers to frames of the video and creates a summary video that includes the background of the video and identified instances for simultaneous playback within the fixed field of view. The media detection system may also identify instances of objects in a live video stream and use the identified instances to respond to user questions. The media detection system applies a language model to questions to identify the subject matter of the questions, identifies content within the live video stream associated with the subject matter, and uses the identified content to respond to the user's question.

Abstract: A media detection system receives a video corresponding to a fixed field of view. The media detection system may receive user input indicating one or more object types to identify or a subset of the video within which to identify objects. The media detection system applies one or more machine-learned classifiers to frames of the video and creates a summary video that includes the background of the video and identified instances for simultaneous playback within the fixed field of view. The media detection system may also identify instances of objects in a live video stream and use the identified instances to respond to user questions. The media detection system applies a language model to questions to identify the subject matter of the questions, identifies content within the live video stream associated with the subject matter, and uses the identified content to respond to the user's question.

Abstract: A media detection system receives a video corresponding to a fixed field of view. The media detection system may receive user input indicating one or more object types to identify or a subset of the video within which to identify objects. The media detection system applies one or more machine-learned classifiers to frames of the video and creates a summary video that includes the background of the video and identified instances for simultaneous playback within the fixed field of view. The media detection system may also identify instances of objects in a live video stream and use the identified instances to respond to user questions. The media detection system applies a language model to questions to identify the subject matter of the questions, identifies content within the live video stream associated with the subject matter, and uses the identified content to respond to the user's question.

Abstract: A media detection system receives a video corresponding to a fixed field of view. The media detection system may receive user input indicating one or more object types to identify or a subset of the video within which to identify objects. The media detection system applies one or more machine-learned classifiers to frames of the video and creates a summary video that includes the background of the video and identified instances for simultaneous playback within the fixed field of view. The media detection system may also identify instances of objects in a live video stream and use the identified instances to respond to user questions. The media detection system applies a language model to questions to identify the subject matter of the questions, identifies content within the live video stream associated with the subject matter, and uses the identified content to respond to the user's question.

Abstract: A method comprising: receiving a radio map of an indoor venue using survey data collected by a survey device positioned throughout the venue, the radio map including a boundary; obtaining harvest data that correspond to locations that are outside of the boundary; filtering the harvest data; and extending the radio map using the survey data and the filtered harvest data, wherein the extended radio map is defined at least in part by an extension of the boundary.

Abstract: Embodiments are disclosed for compressing radio maps of fingerprint-based positioning systems using different compression models. In an embodiment, a method comprises: receiving, by a computing device, access point (AP) data from a plurality of mobile devices operating in a geographic region, the AP data including signal strength measurements of AP signals received at a plurality of reference locations in the geographic region and uncertainty measurements associated with the signal strength measurements; determining a level of accuracy with the first compression model; responsive to the determining, selecting one of the first compression model or a second compression model to compress the AP data, the second compression model being different than the first compression model; compressing the AP data using the selected compression model; and responsive to a request from a mobile device operating in the geographic region, sending a data packet including the compressed AP data to the mobile device.

Abstract: A method comprising: receiving a harvest trace from a mobile device, the harvest trace including a plurality of location fixes each corresponding to a location at a venue, each location fix associated with data including one or more RSSI measurements of one or more wireless signals received from each of a plurality of wireless access points positioned at the venue, the plurality of location fixes including at least two location fixes that have a positional relationship with each other; comparing the data associated with one of the location fixes with data associated with a reference point identified in a radio map of the venue; and updating the radio map of the venue by updating the data associated with the reference point based on the data associated with the one of the location fixes.

Abstract: Embodiments are disclosed for compressing radio maps of fingerprint-based positioning systems using different compression models. In an embodiment, a method comprises: receiving, by a computing device, access point (AP) data from a plurality of mobile devices operating in a geographic region, the AP data including signal strength measurements of AP signals received at a plurality of reference locations in the geographic region and uncertainty measurements associated with the signal strength measurements; determining a level of accuracy with the first compression model; responsive to the determining, selecting one of the first compression model or a second compression model to compress the AP data, the second compression model being different than the first compression model; compressing the AP data using the selected compression model; and responsive to a request from a mobile device operating in the geographic region, sending a data packet including the compressed AP data to the mobile device.

Abstract: Embodiments are disclosed for compressing radio maps of fingerprint-based positioning systems. In an embodiment, a method comprises: receiving access point (AP) data from a plurality of mobile devices operating in a geographic region, the AP data including signal strength measurements of AP signals received at a plurality of reference locations in the geographic region; filtering the AP data to remove outlier AP data; fitting a surface to the AP data; projecting AP data at surface control points onto a two-dimensional image grid; determining a boundary surrounding locations of the AP data at the surface control points; encoding the boundary; encoding the AP data at the surface control points included within the boundary; generating compressed radio maps from the encoded AP data; and responsive to a request from a mobile device operating in the geographic region, sending a data packet including the compressed radio maps to the mobile device.

Abstract: A method comprising: receiving a harvest trace from a mobile device, the harvest trace including a plurality of location fixes each corresponding to a location at a venue, each location fix associated with data including one or more RSSI measurements of one or more wireless signals received from each of a plurality of wireless access points positioned at the venue, the plurality of location fixes including at least two location fixes that have a positional relationship with each other; comparing the data associated with one of the location fixes with data associated with a reference point identified in a radio map of the venue; and updating the radio map of the venue by updating the data associated with the reference point based on the data associated with the one of the location fixes.

Abstract: Embodiments are disclosed for compressing radio maps of fingerprint-based positioning systems. In an embodiment, a method comprises: receiving access point (AP) data from a plurality of mobile devices operating in a geographic region, the AP data including signal strength measurements of AP signals received at a plurality of reference locations in the geographic region; filtering the AP data to remove outlier AP data; fitting a surface to the AP data; projecting AP data at surface control points onto a two-dimensional image grid; determining a boundary surrounding locations of the AP data at the surface control points; encoding the boundary; encoding the AP data at the surface control points included within the boundary; generating compressed radio maps from the encoded AP data; and responsive to a request from a mobile device operating in the geographic region, sending a data packet including the compressed radio maps to the mobile device.

Abstract: A method comprising: receiving a radio map of an indoor venue using survey data collected by a survey device positioned throughout the venue, the radio map including a boundary; obtaining harvest data that correspond to locations that are outside of the boundary; filtering the harvest data; and extending the radio map using the survey data and the filtered harvest data, wherein the extended radio map is defined at least in part by an extension of the boundary.

Abstract: Methods, systems, and computer program product for determining a building floor level are described. A mobile device can use wireless signal sources and location fingerprint data to determine a level of a building floor on which the mobile device is located. The location fingerprint data can include or be associated with a list and a count of wireless signal sources previously detected on each floor. The mobile device can compare the list and count with wireless signal sources detected by the mobile device, and use results of the comparison to configure a statistical filter that determines a location of the mobile device. The mobile device can then determine the location, including a building floor level, using the statistical filter.

Abstract: Methods, systems, and computer program products for correcting in-venue location estimation using structural information are described. A mobile device can use wireless location technologies and dead reckoning to determine an estimated location of the mobile device in a venue. The mobile device can compare the estimated location with a map of the venue. Upon determining that the estimated location conflicts with a structural constraint, the mobile device can adjust the location estimation using the structural information. Adjusting the location estimation can include adjusting a statistical filter that provides estimation of the location and changing a heading of the mobile device used in the dead reckoning.

Abstract: Methods, systems, and computer program product for prefetching location data based on predicted user behavior. A mobile device can request, from a user routine subsystem of the mobile device, a list of locations that a user of the mobile device routinely visits while the user carries the mobile device. The mobile device can determine a cluster of these locations that are within a specified distance between one another. The mobile device can request location data for these locations from a location server, even if the user is not at one of these locations. The location data can include a venue map and a venue location fingerprint. Upon detecting that the user entered a venue at one of these locations, the mobile device can determine a location of the user inside of the venue using the venue location fingerprint. The mobile device can then display the location on a venue map.