Abstract: A navigation system for providing driving instructions to a driver of a vehicle traveling on a route is provided. The driving instructions are generated by executing a multimodal fusion method that comprises extracting features from sensor measurements, annotating the features with directions for the vehicle to follow the route with respect to objects sensed by the sensors, and encoding the annotated features with a multimodal attention neural network to produce encodings. The encodings are transformed into a common latent space, and the transformed encodings are fused using an attention mechanism producing an encoded representation of the scene. The method further comprises decoding the encoded representation with a sentence generation neural network to generate a driving instruction and submitting the driving instruction to an output device.

Abstract: A navigation system configured to provide driving instructions to a driver of a moving vehicle based on real-time description of objects in a scene pertinent to driving the vehicle is provided.

Abstract: A point cloud encoder including an input interface to accept a dynamic point cloud including a sequence of point cloud frames of a scene. A processor encodes blocks of a current point cloud frame to produce an encoded frame. Wherein, for encoding a current block of the current point cloud frame, a reference block is selected similar to the current block according to a similarity metric to serve as a reference to encode the current block. Pair each point in the current block to a point in the reference block based on values of the paired points. Encode the current block based on a combination of an identification of the reference block and residuals between the values of the paired points. Wherein the residuals are ordered according to an order of the values of the points in the reference block. A transmitter transmits the encoded frame over a communication channel.

Abstract: Systems and methods for a point cloud decoder including a processor to decode each block in a set of blocks from a point cloud, so as to obtain a decoded point cloud. Wherein each block includes a set of points, such that for each block the processor is to decode a set of prediction residuals from a compressed bitstream. Use a predetermined location in the block, and compute for each prediction residual in the set of prediction residuals, a position of a point by adding the prediction residual to the predetermined location, so as to obtain a set of decoded points for the block. Wherein the decoded points for the blocks in the set of blocks represent the decoded point cloud.

Abstract: A point cloud encoder including an input interface to accept a dynamic point cloud including a sequence of point cloud frames of a scene. A processor encodes blocks of a current point cloud frame to produce an encoded frame. Wherein, for encoding a current block of the current point cloud frame, a reference block is selected similar to the current block according to a similarity metric to serve as a reference to encode the current block. Pair each point in the current block to a point in the reference block based on values of the paired points. Encode the current block based on a combination of an identification of the reference block and residuals between the values of the paired points. Wherein the residuals are ordered according to an order of the values of the points in the reference block. A transmitter transmits the encoded frame over a communication channel.

Abstract: Systems and methods for determining a pattern in time series data representing an operation of a machine. A memory to store and provide a set of training data examples generated by a sensor of the machine, wherein each training data example represents an operation of the machine for a period of time ending with a failure of the machine. A processor configured to iteratively partition each training data example into a normal region and an abnormal region, determine a predictive pattern absent from the normal regions and present in each abnormal region only once, and determine a length of the abnormal region. Outputting the predictive pattern via an output interface in communication with the processor or storing the predictive pattern in memory, wherein the predictive pattern is a predictive estimate of an impending failure and assists in management of the machine.

Abstract: A set of input images are acquired sequentially as image tensors. A low-tubal rank tensor and a sparse tensor are initialized using the image tensor, wherein the low-tubal rank tensor is a tensor product of a low-rank spanning tensor basis and corresponding tensor coefficients, and for each image, updating iteratively the image tensor, the tensor coefficients, and the sparse tensor using the image tensor and the low-rank spanning basis from a previous iteration. The spanning tensor basis is updated using the tensor coefficients, the sparse tensor, and the low rank tubal tensor, wherein the low rank tubal tensor represents a set of output images and the sparse tensor representing a set of sparse images.

Abstract: Systems and methods for image processing including a memory having stored multi-angled view images of a scene. Each multi-angled view image includes pixels, and at least one multi-angled view image includes a clouded area in the scene, resulting in missing pixels. A processor configured to align three multi-angled view images in the multi-angled view images to a target view angle of the scene, to form a set of aligned multi-angled view images representing a target point of view of the scene, at least one aligned multi-angled view image of the at least three multi-angled view images, has missing pixels due to the clouded area. Form a matrix using vectorized aligned multi-angled view images, wherein the matrix is incomplete due to the missing pixels. Complete the matrix using a matrix completion to combine the aligned multi-angled view images to produce a fused image of the scene without the clouded area.

Abstract: A method processes keypoint trajectories in a video, wherein the keypoint trajectories describe motion of a plurality of keypoints across pictures of the video over time, by first acquiring the video of a scene using a camera. Keypoints and associated feature descriptors are detected in each picture. The keypoints and associated features descriptors are matched between neighboring pictures to generate keypoint trajectories. Then, the keypoint trajectories are coded predictively into a bitstream, which is outputted.

Abstract: Systems and methods for image processing including a memory having stored multi-angled view images of a scene. Each multi-angled view image includes pixels, and at least one multi-angled view image includes a clouded area in the scene, resulting in missing pixels. A processor configured to align three multi-angled view images in the multi-angled view images to a target view angle of the scene, to form a set of aligned multi-angled view images representing a target point of view of the scene, at least one aligned multi-angled view image of the at least three multi-angled view images, has missing pixels due to the clouded area. Form a matrix using vectorized aligned multi-angled view images, wherein the matrix is incomplete due to the missing pixels. Complete the matrix using a matrix completion to combine the aligned multi-angled view images to produce a fused image of the scene without the clouded area.

Abstract: Systems and methods for determining a pattern in time series data representing an operation of a machine. A memory to store and provide a set of training data examples generated by a sensor of the machine, wherein each training data example represents an operation of the machine for a period of time ending with a failure of the machine. A processor configured to iteratively partition each training data example into a normal region and an abnormal region, determine a predictive pattern absent from the normal regions and present in each abnormal region only once, and determine a length of the abnormal region. Outputting the predictive pattern via an output interface in communication with the processor or storing the predictive pattern in memory, wherein the predictive pattern is a predictive estimate of an impending failure and assists in management of the machine.

Abstract: A method for encoding a point cloud representing a scene using an encoder including a processor in communication with a memory includes steps of fitting a parameterized surface onto the point cloud having input points representing locations in a three-dimensional space, generating model parameters from the parameterized surface, computing corresponding points from the parameterized surface, wherein the corresponding points correspond to the input points, computing residual data based on the corresponding points and the input points of the point cloud, compressing the model parameters and residual data to yield coded model parameters and coded residual data, respectively, and producing a bit-stream from the coded model parameters of the parameterized surface and the coded residual data.

Abstract: A method and system determines flows by first acquiring a video of the flows with a camera, wherein the flows are pedestrians in a scene, wherein the video includes a set of frames. Motion vectors are extracted from each frame in the set, and a data matrix is constructed from the motion vectors in the set of frames. A low rank Koopman operator is determined from the data matrix and a spectrum of the low rank Koopman operator is analyzed to determine a set of Koopman modes. Then, the frames are segmented into independent flows according to a clustering of the Koopman modes.

Abstract: A method processes keypoint trajectories in a video, wherein the keypoint trajectories describe motion of a plurality of keypoints across pictures of the video over time, by first acquiring the video of a scene using a camera. Kkeypoints and associated feature descriptors are detected in each picture. The keypoints and associated features descriptors are matched between neighboring pictures to generate keypoint trajectories. Then, the keypoint trajectories are coded predictively into a bitstream, which is outputted.

Abstract: A method compresses a point cloud composed of a plurality of points in a three-dimensional (3D) space by first acquiring the point cloud with a sensor, wherein each point is associated with a 3D coordinate and at least one attribute. The point cloud is partitioned into an array of 3D blocks of elements, wherein some of the elements in the 3D blocks have missing points. For each 3D block, attribute values for the 3D block are predicted based on the attribute values of neighboring 3D blocks, resulting in a 3D residual block. A 3D transform is applied to each 3D residual block using locations of occupied elements to produce transform coefficients, wherein the transform coefficients have a magnitude and sign. The transform coefficients are entropy encoded according the magnitudes and sign bits to produce a bitstream.

Abstract: A method separates foreground from background in a sequence of images, by first acquiring the sequence of images and a depth map of a scene by a camera. Groups of pixels are determined based on the depth map. Then, the sequence of images is decomposed into a sparse foreground component, and a low rank background component, according to apparent motion in the sequence of images, and the groups.

Abstract: A method and system determines flows by first acquiring a video of the flows with a camera, wherein the flows are pedestrians in a scene, wherein the video includes a set of frames. Motion vectors are extracted from each frame in the set, and a data matrix is constructed from the motion vectors in the set of frames. A low rank Koopman operator is determined from the data matrix and a spectrum of the low rank Koopman operator is analyzed to determine a set of Koopman modes. Then, the frames are segmented into independent flows according to a clustering of the Koopman modes.

Abstract: A method extracts a low-rank descriptor of a video acquired of a scene by first extracting a set of descriptors for each image in the video. The sets of descriptors for the video are aggregated to form a descriptor matrix. Iteratively, a low-rank descriptor matrix is determined from the descriptor matrix, as well as a selection matrix that associates each column in the descriptor matrix to a corresponding column in the low-rank descriptor matrix. The low-rank descriptor matrix is output upon convergence.

Abstract: A method processes a signal represented as a graph by first determining a graph spectral transform based on the graph. In a spectral domain, parameters of a graph filter are estimated using a training data set of unenhanced and corresponding enhanced signals. The graph filter is derived based on the graph spectral transform and the estimated graph filter parameters. Then, the signal is processed using the graph filter to produce an output signal. The processing can enhance signals such as images by denoising or interpolating missing samples.

Abstract: A method processes a signal by first constructing a graph from the signal, and then determining a graph matrix from the graph and the signal. A Krylov-based subspace is determined based on the graph matrix and the signal. A filter for the Krylov subspace is determined. The filter transforms the signal to produce a filtered signal, which is output.