Abstract: A method includes receiving an input comprising multi-modal inputs such as text, audio, video, or context information from a client device associated with a user, assigning a task associated with the input to a server among a plurality of servers, determining a context response corresponding to the input based on the input and interaction history between the computing system and the user, generating meta data specifying expressions, emotions, and non-verbal and verbal gestures associated with the context response by querying a trained behavior knowledge graph, generating media content output based on the determined context response and the generated meta data, the media content output comprising of text, audio, and visual information corresponding to the determined context response in the expressions, the emotions, and the non-verbal and verbal gestures specified by the meta data, sending instructions for presenting the generated media content output to the user to the client device.

Abstract: A system and method for a master platform includes receiving first pose data associated with an image sensor of a first device, and a first semantic map generated by the first device, the first semantic map including a simplified object representation in a coordinate space of the first device. The master platform also receives second pose data associated with an image sensor of a second device, and a second semantic map generated by the second device, the second semantic map including a simplified object representation in a coordinate space of the second device. A shared simplified object representation common to the first and semantic maps is identified. The master platform further combines the first semantic and second semantic maps based on the first and second pose data. The first pose data, first semantic map, second pose data, and second semantic map are associated with a common time interval.

Abstract: An untethered apparatus for performing inside-out device tracking based on visual-inertial simultaneous location and mapping (SLAM) includes a dynamic vision sensor (DVS) configured to output an asynchronous stream of sensor event data, an inertial measurement unit (IMU) sensor configured to collect IMU data associated with motion of the apparatus at a predetermined interval, a processor and a memory. The memory contains instructions, which when executed by the processor, cause the apparatus to accumulate DVS sensor output over a sliding time window, the sliding time window including the predetermined interval, apply a motion correction to the accumulated DVS sensor output, the motion correction based on the IMU data collected over the predetermined interval, generate an event-frame histogram of DVS sensor events based on the motion correction, and provide the event-frame histogram of the DVS sensor events and the IMU data to a visual inertial SLAM pipeline.

Abstract: A computer-implemented method of providing an emotion-aware reactive interface in an electronic device includes receiving an image of a user as an input and identifying a multi-modal non-verbal cue in the image. The method further includes interpreting the multi-modal non-verbal cue to determine a categorization and outputting a reactive interface event determined based on the categorization.

Abstract: An apparatus includes a dynamic vision sensor (DVS) configured to output an asynchronous stream of sensor event data, and a complementary metal-oxide-semiconductor (CMOS) image sensor configured to output frames of image data. The apparatus further includes a hybrid feature handler configured to receive, as an input, one or more of a DVS output or a CMOS image sensor output, and provide tracked features to a visual-inertial simultaneous location and mapping (SLAM) pipeline performing inside-out device tracking, and a sensor scheduler configured to switch off the CMOS image sensor based on a current value of one or more CMOS control factors.

Abstract: A method for semi-dense depth estimation includes receiving, at an electronic device, a control signal of a speckle pattern projector (SPP and receiving from each sensor of a dynamic vision sensor (DVS) stereo pair, an event stream of pixel intensity change data, wherein the event stream is time-synchronized with the control signal of the SPP. The method further includes performing projected light filtering on the event stream of pixel intensity change data for each sensor of the DVS stereo pair, to generate synthesized event image data, the synthesized event image data having one or more channels, each channel based on an isolated portion of the event stream of pixel intensity change data and performing stereo matching on at least one channel of the synthesized event image data for each sensor of the DVS stereo pair to generate a depth map for at least a portion of the field of view.

Abstract: An untethered apparatus for performing inside-out device tracking based on visual-inertial simultaneous location and mapping (SLAM) includes a dynamic vision sensor (DVS) configured to output an asynchronous stream of sensor event data, an inertial measurement unit (IMU) sensor configured to collect IMU data associated with motion of the apparatus at a predetermined interval, a processor and a memory. The memory contains instructions, which when executed by the processor, cause the apparatus to accumulate DVS sensor output over a sliding time window, the sliding time window including the predetermined interval, apply a motion correction to the accumulated DVS sensor output, the motion correction based on the IMU data collected over the predetermined interval, generate an event-frame histogram of DVS sensor events based on the motion correction, and provide the event-frame histogram of the DVS sensor events and the IMU data to a visual inertial SLAM pipeline.

Abstract: A method includes retrieving, by a device, contextual information based on at least one of an image, the device, user context, or a combination thereof. At least one model is identified from multiple models based on the contextual information and at least one object recognized in an image based on at least one model. At least one icon is displayed at the device. The at least one icon being associated with at least one of an application, a service, or a combination thereof providing additional information.

Abstract: One embodiment provides a method comprising classifying one or more objects present in an input comprising visual data by executing a first set of models associated with a domain on the input. Each model corresponds to an object category. Each model is trained to generate a visual classifier result relating to a corresponding object category in the input with an associated confidence value indicative of accuracy of the visual classifier result. The method further comprises aggregating a first set of visual classifier results based on confidence value associated with each visual classifier result of each model of the first set of models. At least one other model is selectable for execution on the input based on the aggregated first set of visual classifier results for additional classification of the objects. One or more visual classifier results are returned to an application running on an electronic device for display.

Abstract: A system and method for a master platform includes receiving first pose data associated with an image sensor of a first device, and a first semantic map generated by the first device, the first semantic map including a simplified object representation in a coordinate space of the first device. The master platform also receives second pose data associated with an image sensor of a second device, and a second semantic map generated by the second device, the second semantic map including a simplified object representation in a coordinate space of the second device. A shared simplified object representation common to the first and semantic maps is identified. The master platform further combines the first semantic and second semantic maps based on the first and second pose data. The first pose data, first semantic map, second pose data, and second semantic map are associated with a common time interval.

Abstract: An apparatus includes a dynamic vision sensor (DVS) configured to output an asynchronous stream of sensor event data, a CMOS image sensor configured to output frames of image data, an inertial measurement unit (IMU), a processor and a memory. The memory contains instructions, which when executed by the processor, cause the apparatus to generate a semantic segmentation of a time-stamped frame, which is based on one or more of an output of the CMOS image sensor, or a synthesized event frame based on an output from the DVS and an output from the IMU over a time interval. The semantic segmentation includes a semantic label associated with a region of the time-stamped frame. When executed, the instructions further cause the apparatus to determine, based on the semantic segmentation, a simplified object representation in a coordinate space, and update a stable semantic map based on the simplified object representation.

Abstract: In one embodiment, a computing system may access one or more frames of media content with each frame corresponding to at least part of an image. The computing system may automatically detect at least one object in at least one of the one or more frames. The computing system may automatically identify the at least one object. For each of the identified objects, the computing system may automatically determine an annotation corresponding to that object and each annotation may include one or more graphical elements. For each annotation, the computing system may automatically associate that annotation for display with its corresponding object.

Abstract: A method for semi-dense depth estimation includes receiving, at an electronic device, a control signal of a speckle pattern projector (SPP and receiving from each sensor of a dynamic vision sensor (DVS) stereo pair, an event stream of pixel intensity change data, wherein the event stream is time-synchronized with the control signal of the SPP. The method further includes performing projected light filtering on the event stream of pixel intensity change data for each sensor of the DVS stereo pair, to generate synthesized event image data, the synthesized event image data having one or more channels, each channel based on an isolated portion of the event stream of pixel intensity change data and performing stereo matching on at least one channel of the synthesized event image data for each sensor of the DVS stereo pair to generate a depth map for at least a portion of the field of view.

Abstract: An apparatus includes a dynamic vision sensor (DVS) configured to output an asynchronous stream of sensor event data, a CMOS image sensor configured to output frames of image data, an inertial measurement unit (IMU), a processor and a memory. The memory contains instructions, which when executed by the processor, cause the apparatus to generate a semantic segmentation of a time-stamped frame, which is based on one or more of an output of the CMOS image sensor, or a synthesized event frame based on an output from the DVS and an output from the IMU over a time interval. The semantic segmentation includes a semantic label associated with a region of the time-stamped frame. When executed, the instructions further cause the apparatus to determine, based on the semantic segmentation, a simplified object representation in a coordinate space, and update a stable semantic map based on the simplified object representation.

Abstract: An apparatus includes a dynamic vision sensor (DVS) configured to output an asynchronous stream of sensor event data, and a complementary metal-oxide-semiconductor (CMOS) image sensor configured to output frames of image data. The apparatus further includes a hybrid feature handler configured to receive, as an input, one or more of a DVS output or a CMOS image sensor output, and provide tracked features to a visual-inertial simultaneous location and mapping (SLAM) pipeline performing inside-out device tracking, and a sensor scheduler configured to switch off the CMOS image sensor based on a current value of one or more CMOS control factors.

Abstract: A computer-implemented method of providing an emotion-aware reactive interface in an electronic device includes receiving an image of a user as an input and identifying a multi-modal non-verbal cue in the image. The method further includes interpreting the multi-modal non-verbal cue to determine a categorization and outputting a reactive interface event determined based on the categorization.

Abstract: In one embodiment, a computing system may access one or more frames of media content with each frame corresponding to at least part of an image. The computing system may automatically detect at least one object in at least one of the one or more frames. The computing system may automatically identify the at least one object. For each of the identified objects, the computing system may automatically determine an annotation corresponding to that object and each annotation may include one or more graphical elements. For each annotation, the computing system may automatically associate that annotation for display with its corresponding object.

Abstract: One embodiment provides a method comprising classifying one or more objects present in an input comprising visual data by executing a first set of models associated with a domain on the input. Each model corresponds to an object category. Each model is trained to generate a visual classifier result relating to a corresponding object category in the input with an associated confidence value indicative of accuracy of the visual classifier result. The method further comprises aggregating a first set of visual classifier results based on confidence value associated with each visual classifier result of each model of the first set of models. At least one other model is selectable for execution on the input based on the aggregated first set of visual classifier results for additional classification of the objects. One or more visual classifier results are returned to an application running on an electronic device for display.

Abstract: A method includes retrieving, by a device, contextual information based on at least one of an image, the device, user context, or a combination thereof. At least one model is identified from multiple models based on the contextual information and at least one object recognized in an image based on at least one model. At least one icon is displayed at the device. The at least one icon being associated with at least one of an application, a service, or a combination thereof providing additional information.