Abstract: In one embodiment, a method includes accessing an initial image and applying each image filter from a set of one or more N image filters to the initial image to create N filtered images. The method further includes generating a similarity metric for each pair of images in a set of images comprising the initial image and the N filtered images and determining, based on the similarity metrics, a placement of each image in the set of images within an image-filter space. The method further includes generating, based on the image-filter space, a selectable image-filter interface; and providing the selectable image-filter interface for display, wherein each selectable region of the selectable image-filter interface is associated with a particular filtering of the initial image.

Abstract: Image-to-image translations using 1D inputs includes concatenating multiple 1D vectors forming a concatenated 1D vector. The multiplicity of 1D vectors includes 1D vectors of at least two different modalities. An encoded 1D vector is generated by encoding the concatenated 1D vector. An encoded 2D array of features is generated by reshaping an arrangement of features of the encoded 1D feature vector. An image of a virtual human is generated by decoding the encoded 2D array.

Abstract: Synthesizing speech and movement of a virtual human includes capturing supplemental data generated by a transducer. The supplemental data specifies one or more attributes of a user. The capturing is performed in substantially real-time with the user providing input to a conversational platform. A behavior determiner generates behavioral data based on the supplemental data and an audio response generated by the conversational platform in response to the input to the conversation platform. Based on the behavioral data and the audio response, a rendering network generates a video rendering of a virtual human engaging in a conversation with the user, the video rendering synchronized with the audio response.

Abstract: In one embodiment, a method includes generating, from an accessed image, one or more image patches, and for each image patch: (1) accessing a set of point-spread functions (PSFs), wherein each PSF in the set of PSFs corresponds to one of a plurality of points in the image patch; (2) generating a set of deconvolved image patches by deconvolving the image patch with each PSF from the set of PSFs for that patch; (3) determining, for each of one or more portions of the image patch, a set of weights, wherein each weight in the set of weights is associated with one of the deconvolved image patches; and (4) generating a restored image patch by interpolating the set of deconvolved image patches based on the set of weights.

Abstract: In one embodiment, a method includes accessing an image captured by an imaging system that includes an aperture and an optical sensor, wherein the imaging system blurs the image; and de-blurring the image by applying an Airy-disk correction that attributes at least part of the blurring in the image to a circular aperture rather than to a pinhole aperture.

Abstract: Radar-based sensing includes, at an electronic device that includes one or more radar-based transmitters and one or more radar-based receivers, transmitting, from one or more of the radar-based transmitters, radio waves into an area in a vicinity of the electronic device and receiving, at one or more of the radar-based receivers, radio waves reflected off one or more objects in the vicinity of the electronic device. The radio waves, as reflected off the one or more objects, can be processed using a radar imaging module.

Abstract: A manipulator may include a base link having a ring shape; a first ring-shaped link having a ring shape and disposed to rotate at a predetermined angle at a front end of the base link; a second ring-shaped link having a ring shape and disposed to rotate at a predetermined angle at a front end of the first ring-shaped link; a first Borden cable and a second Borden cable connected to the first ring-shaped link and formed to rotate the first ring-shaped link; a third Borden cable and a fourth Borden cable connected to the second ring-shaped link and formed to rotate the second ring-shaped link; and a driving device that operates the first Borden cable, the second Borden cable, the third Borden cable, and the fourth Borden cable.

Abstract: Multimodal disentanglement can include generating a set of silhouette images corresponding to a human face, the generating undoing a correlation between an upper portion and a lower portion of the human face depicted by each silhouette image. A unimodal machine learning model can be trained with the set of silhouette images. As trained, the unimodal machine learning model can generate synthetic images of the human face. The synthetic images generated by the unimodal machine learning model once trained can be used to train a multimodal rendering network. The multimodal rendering network can be trained to generate a voice-animated digital human. Training the multimodal rendering network can be based on minimizing differences between the synthetic images and images generated by the multimodal rendering network.

Abstract: An apparatus for a robotic limb includes one or more limb segments connected via one or more joints. The robotic limb may feature one or more dual-reduction quasi-quasi-direct-drive joint actuators that permit the robotic limb to move throughout a scene. The robotic limb may further include an end-effector connected to a free end of the robotic limb with one or more opposable fingers comprising a four bar linkage. The end-effector may include a main actuator that actuates the one or more fingers via the four-bar linkages to complete various tasks.

Abstract: Creating images and animations of lip motion from mouth shape data includes providing, as one or more input features to a neural network model, a vector of a plurality of coefficients. Each vector of the plurality of coefficients corresponds to a different mouth shape. Using the neural network model, a data structure output specifying a visual representation of a mouth including lips having a shape corresponding to the vector is generated.

Abstract: Image generation using a hierarchical, model-based system includes generating a first region of an image using a first neural network model. The first region of the image is provided to a second neural network model as input. A second region of the image is generated using the second neural network model. The second region of the image shares a boundary with at least a portion of the first region of the image.

Abstract: An apparatus and method for a wall clock AI voice assistant, such as a computing device, is provided herein. The wall-mountable electronic clock includes an optical sensor, a microphone, a digital display and a speaker. The optical sensor captures image data representing at least a portion of a user. The microphone receives an audio input from the user when certain criteria is met, such as when the image data representing at least a portion of the user is captured. The digital display provides visual information to the user provided in response to the captured image data or the audio input. The speaker provides audio information provides from a digital voice assistant in response to the captured image data or the audio input.

Abstract: In one embodiment, a method of training a machine-learning model for modifying a facial illumination in an image includes accessing an initial image including a human face having an initial illumination and determining one or more illumination priors for the initial image. The method includes providing the initial image and the one or more illumination priors to the machine-learning model; receiving, from the machine-learning model, a set of correction operators identifying a modified illumination for the human face; creating, based at least on the set of correction operators and the initial image, a modified image having the modified illumination; creating, based on the modified image, a reconstructed initial image including the human face having a reconstructed illumination; and adjusting one or more parameters of the machine-learning model by minimizing a loss function based on a difference between the initial and the reconstructed initial images in their respective illumination.

Abstract: A manipulator and a method for controlling the manipulator are disclosed.

Abstract: Image de-skewing can include acquiring pixel coordinates of an image captured by a camera adjoining an electronic display device. The pixel coordinates can be mapped to de-skewing coordinates using a predetermined de-skewing transformer that maps the pixel coordinates to de-skewing coordinates. A de-skewed image can be generated based on the de-skewing coordinates such that the de-skewed image is perpendicular to a principal axis extending from a predetermined location of an apparent camera on the electronic display device.

Abstract: In one embodiment, a health-monitoring system may access a waist-hip measurement of a user. The system may determine one or more stress-related parameters of the user using one or more computing devices. The system may determine one or more correlations between the waist-hip measurement and the one or more stress-related parameters of the user. The system may provide feedback to the user based on one or more of the one or more stress-related parameters or the determined correlations between the waist-hip measurement and the one or more stress-related parameters.

Abstract: A method includes capturing, by a camera disposed behind a display panel of an electronic device, an original image through a semi-transparent pixel region of the display panel. The original image includes one or more color components. The method further includes determining, for a plurality of pixel regions of the original image, a point spread function (PSF) for each of the one or more color components. The method further includes performing, for the plurality of pixel regions of the original image, a deconvolution of each of the one or more color components of the original image based at least in part on their respective PSFs. The method thus includes generating a reconstructed image corresponding to the original image based on the deconvolutions of the one or more color components of the plurality of pixel regions of the original image.

Abstract: A method by an electromagnetic device includes determining a pattern of electromagnetic energy absorbed by a load disposed inside a cavity into which electromagnetic radiation is directed and generating one or more maps of the pattern of electromagnetic energy absorbed by the load. The one or more maps comprises an indication of a distribution of heating within the load. The method further includes determining, based on the one or more maps, a plurality of sequences of operating parameter combinations configured so as to heat the load via absorption of the electromagnetic radiation in accordance with a target temperature profile with respect to the load. The method thus includes emitting electromagnetic radiation into the cavity based on the plurality of sequences of operating parameter combinations to achieve the target temperature profile with respect to the load.

Abstract: In one embodiment, a method includes displaying functions associated with a remote device at a wearable computing device, wherein the remote device is paired with the wearable computing device, receiving a user input from a user at the wearable computing device, determining intended operations of one or more of the functions by the user based on the user input, sending instructions for executing the intended operations of the one or more of the functions associated with the remote device to the remote device, and displaying execution results of the intended operations of the one or more of the functions associated with the remote device at the wearable computing device.

Abstract: Radar-based sensing includes, at an electronic device that includes one or more radar-based transmitters and one or more radar-based receivers, transmitting, from one or more of the radar-based transmitters, radio waves into an area in a vicinity of the electronic device and receiving, at one or more of the radar-based receivers, radio waves reflected off one or more objects in the vicinity of the electronic device. The radio waves, as reflected off the one or more objects, can be processed using a radar imaging module.