Abstract: Estimating a three-dimensional (3D) pose of an object, such as a hand or body (human, animal, robot, etc.), from a 2D image is necessary for human-computer interaction. A hand pose can be represented by a set of points in 3D space, called keypoints. Two coordinates (x,y) represent spatial displacement and a third coordinate represents a depth of every point with respect to the camera. A monocular camera is used to capture an image of the 3D pose, but does not capture depth information. A neural network architecture is configured to generate a depth value for each keypoint in the captured image, even when portions of the pose are occluded, or the orientation of the object is ambiguous. Generation of the depth values enables estimation of the 3D pose of the object.

Abstract: A client and content provider are connected by a plurality of simultaneous transport connections. The number of the transport connections that are used to transfer data is selected based on the size of the data to be transferred and may change after transfer of data has commenced based on the amount of data left and the attributes of the transport connections. In another aspect, data to be transmitted over the transport connections is organized into frames such that each frame includes data from only one data stream. The frames are sized to be less than or equal to a control window of the transport connection over which they are transmitted. Each frame may be assigned to a transport connection in a round robin fashion or based on the size of the frame and the sizes of the control windows of the transport connections.

Abstract: A method, computer readable medium, and system are disclosed for classifying video image data. The method includes the steps of processing training video image data by at least a first layer of a convolutional neural network (CNN) to extract a first set of feature maps and generate classification output data for the training video image data. Spatial classification accuracy data is computed based on the classification output data and target classification output data and spatial discrimination factors for the first layer are computed based on the spatial classification accuracies and the first set of feature maps.

Abstract: Some embodiments include a plasma source. The plasma source includes: (i) a plasma emitter having a first electrode and a second electrode defining an electrode gap therebetween; (ii) stands disposed adjacent to the electrode gap and the plasma emitter; (iii) emitter openings configured such that shockwaves generated by the plasma source are directed through the emitter openings and radially from the plasma emitter, wherein adjacent emitter openings of the emitter openings are separated from each other by at least one stand of the stands; (iv) an enclosure housing at a distal end of the plasma emitter and having a delivery device configured to introduce a conductor through an opening in the second electrode and into the electrode gap; and a device housing at a proximal end of the plasma emitter and having a transformer, a capacitor unit, and a contactor. Other embodiments of related systems and methods are also disclosed.

Abstract: Various embodiments enable a robot, or other autonomous or semi-autonomous device or system, to receive data involving the performance of a task in the physical world. The data can be provided as input to a perception network to infer a set of percepts about the task, which can correspond to relationships between objects observed during the performance. The percepts can be provided as input to a plan generation network, which can infer a set of actions as part of a plan. Each action can correspond to one of the observed relationships. The plan can be reviewed and any corrections made, either manually or through another demonstration of the task. Once the plan is verified as correct, the plan (and any related data) can be provided as input to an execution network that can infer instructions to cause the robot, and/or another robot, to perform the task.

Abstract: A virtual private router (VPR) intercepts DNS requests and returns a pseudo IP address to the requesting application and the pseudo IP address is mapped to a domain name in the request. Requests for content including the pseudo IP address are modified to include the corresponding domain name and transmitted to an intermediary server, which resolves the domain name to a real IP address and forwards the content request. The content is received by the intermediary server, which returns it to the requesting application, such as by way of the VPR. Real IP addresses may be returned by the intermediary server such that subsequent content requests to the domain name may bypass the intermediary server. Requests for certain domains, ports, and/or protocols may bypass the intermediary server such that the VPR resolves the domain names to real IP addresses.

Abstract: An application using a VPN is programmed to transmit proxy traffic to a remote proxy server. Traffic to the proxy server is intercepted, shifted to user space, and processed according to one or more options. Traffic may be terminated by a local proxy that resolves domain names in traffic and requests referenced content. Intercepted traffic may include plain text data in headers that is encrypted before forwarding to a different proxy server. Traffic may be evaluated, such as a User Agent string in order to determine routing choices, such as blocking, throttling, local termination, transmitting through a VPN, or other options. Multiple VPNs may operate on the same user computer and proxy traffic may be intercepted and processed by transmitting it through a VPN, bypassing all VPNs, or routing through a different VPN.

Abstract: An application using a VPN is programmed to transmit proxy traffic to a remote proxy server. Traffic to the proxy server is intercepted, shifted to user space, and processed according to one or more options. Traffic may be terminated by a local proxy that resolves domain names in traffic and requests referenced content. Intercepted traffic may include plain text data in headers that is encrypted before forwarding to a different proxy server. Traffic may be evaluated, such as a User Agent string in order to determine routing choices, such as blocking, throttling, local termination, transmitting through a VPN, or other options. Multiple VPNs may operate on the same user computer and proxy traffic may be intercepted and processed by transmitting it through a VPN, bypassing all VPNs, or routing through a different VPN.

Abstract: An application using a VPN is programmed to transmit proxy traffic to a remote proxy server. Traffic to the proxy server is intercepted, shifted to user space, and processed according to one or more options. Traffic may be terminated by a local proxy that resolves domain names in traffic and requests referenced content. Intercepted traffic may include plain text data in headers that is encrypted before forwarding to a different proxy server. Traffic may be evaluated, such as a User Agent string in order to determine routing choices, such as blocking, throttling, local termination, transmitting through a VPN, or other options. Multiple VPNs may operate on the same user computer and proxy traffic may be intercepted and processed by transmitting it through a VPN, bypassing all VPNs, or routing through a different VPN.

Abstract: Detection of activity in video content, and more particularly detecting in video start and end frames inclusive of an activity and a classification for the activity, is fundamental for video analytics including categorizing, searching, indexing, segmentation, and retrieval of videos. Existing activity detection processes rely on a large set of features and classifiers that exhaustively run over every time step of a video at multiple temporal scales, or as a small improvement computationally propose segments of the video on which to perform classification. These existing activity detection processes, however, are computationally expensive, particularly when trying to achieve activity detection accuracy, and moreover are not configurable for any particular time or computation budget. The present disclosure provides a time and/or computation budget-aware method for detecting activity in video that relies on a recurrent neural network implementing a learned policy.

Abstract: In some embodiments, a plasma source can comprise a plasma emitter comprising a first electrode and a second electrode, the first electrode and the second electrode defining an electrode gap. In some embodiments, the plasma source can further comprise an enclosure housing attached to a distal end of the plasma emitter, the enclosure housing can comprise a delivery device configured to introduce a metal conductor through an axial opening in the second electrode and into the electrode gap and a device housing attached to a proximal end of the plasma emitter, the device housing can comprise a high voltage transformer electrically coupled to a capacitor unit. In some embodiments, the capacitor unit electrically can be coupled to a contactor, and the contactor can be electrically coupled to the first electrode. Other embodiments of related methods and systems are also provided.

Abstract: A client and content provider are connected by a plurality of simultaneous transport connections. The number of the transport connections that are used to transfer data is selected based on the size of the data to be transferred and may change after transfer of data has commenced based on the amount of data left and the attributes of the transport connections. In another aspect, data to be transmitted over the transport connections is organized into frames such that each frame includes data from only one data stream. The frames are sized to be less than or equal to a control window of the transport connection over which they are transmitted. Each frame may be assigned to a transport connection in a round robin fashion or based on the size of the frame and the sizes of the control windows of the transport connections.

Abstract: A virtual private router (VPR) intercepts DNS requests and returns a pseudo IP address to the requesting application and the pseudo IP address is mapped to a domain name in the request. Requests for content including the pseudo IP address are modified to include the corresponding domain name and transmitted to an intermediary server, which resolves the domain name to a real IP address and forwards the content request. The content is received by the intermediary server, which returns it to the requesting application, such as by way of the VPR. Real IP addresses may be returned by the intermediary server such that subsequent content requests to the domain name may bypass the intermediary server. Content requests may be sent to the intermediary server, which may instruct the VPR to bypass the server when bypass is needed.

Abstract: A client and content provider are connected by a plurality of simultaneous transport connections. The number of the transport connections that are used to transfer data is selected based on the size of the data to be transferred and may change after transfer of data has commenced based on the amount of data left and the attributes of the transport connections. In another aspect, data to be transmitted over the transport connections is organized into frames such that each frame includes data from only one data stream. The frames are sized to be less than or equal to a control window of the transport connection over which they are transmitted. Each frame may be assigned to a transport connection in a round robin fashion or based on the size of the frame and the sizes of the control windows of the transport connections.

Abstract: An apparatus and method for gesture detection and recognition. The apparatus includes a processing element, a radar sensor, a depth sensor, and an optical sensor. The radar sensor, the depth sensor, and the optical sensor are coupled to the processing element, and the radar sensor, the depth sensor, and the optical sensor are configured for short range gesture detection and recognition. The processing element is further configured to detect and recognize a hand gesture based on data acquired with the radar sensor, the depth sensor, and the optical sensor.

Abstract: A method, computer readable medium, and system are disclosed for visual sequence learning using neural networks. The method includes the steps of replacing a non-recurrent layer within a trained convolutional neural network model with a recurrent layer to produce a visual sequence learning neural network model and transforming feedforward weights for the non-recurrent layer into input-to-hidden weights of the recurrent layer to produce a transformed recurrent layer. The method also includes the steps of setting hidden-to-hidden weights of the recurrent layer to initial values and processing video image data by the visual sequence learning neural network model to generate classification or regression output data.

Abstract: A method, computer readable medium, and system are disclosed for sequential multi-tasking to generate coordinates of landmarks within images. The landmark locations may be identified on an image of a human face and used for emotion recognition, face identity verification, eye gaze tracking, pose estimation, etc. A neural network model processes input image data to generate pixel-level likelihood estimates for landmarks in the input image data and a soft-argmax function computes predicted coordinates of each landmark based on the pixel-level likelihood estimates.

Abstract: A method, computer readable medium, and system are disclosed to generate coordinates of landmarks within images. The landmark locations may be identified on an image of a human face and used for emotion recognition, face identity verification, eye gaze tracking, pose estimation, etc. A transform is applied to input image data to produce transformed input image data. The transform is also applied to predicted coordinates for landmarks of the input image data to produce transformed predicted coordinates. A neural network model processes the transformed input image data to generate additional landmarks of the transformed input image data and additional predicted coordinates for each one of the additional landmarks. Parameters of the neural network model are updated to reduce differences between the transformed predicted coordinates and the additional predicted coordinates.

Abstract: A method, computer readable medium, and system are disclosed for detecting and classifying hand gestures. The method includes the steps of receiving an unsegmented stream of data associated with a hand gesture, extracting spatio-temporal features from the unsegmented stream by a three-dimensional convolutional neural network (3DCNN), and producing a class label for the hand gesture based on the spatio-temporal features.

Abstract: An apparatus and method for gesture detection and recognition. The apparatus includes a processing element, a radar sensor, a depth sensor, and an optical sensor. The radar sensor, the depth sensor, and the optical sensor are coupled to the processing element, and the radar sensor, the depth sensor, and the optical sensor are configured for short range gesture detection and recognition. The processing element is further configured to detect and recognize a hand gesture based on data acquired with the radar sensor, the depth sensor, and the optical sensor.