Abstract: A method of jointly training of a transferable feature extractor network, an ordinal regressor network, and an order classifier network in an ordinal regression unsupervised domain adaption network by providing a source of labeled source images and unlabeled target images; outputting image representations from a transferable feature extractor network by performing a minimax optimization procedure on the source of labeled source images and unlabeled target images; training a domain discriminator network, using the image representations from the transferable feature extractor network, to distinguish between source images and target images; training an ordinal regressor network using a full set of source images from the transferable feature extractor network; and training an order classifier network using a full set of source images from said transferable feature extractor network.

Abstract: A navigating robot includes: a feature module configured to detect objects in images captured by a camera of the navigating robot while the navigating robot is in a real world space; a mapping module configured to generate a map including locations of objects captured in the images and at least one attribute of the objects; and a navigation module trained to find and navigate to N different objects in the real world space in a predetermined order by: when a location of a next one of the N different objects in the predetermined order is stored in the map, navigate toward the next one of the N different objects in the real world space; and when the location of the next one of the N different objects in the predetermined order is not stored in the map, navigate to a portion of the map not yet captured in any images.

Abstract: A training system includes: a model; and a training module configured to: construct a first pair of images of at least a first portion of a first human captured at different times; construct a second pair of images of at least a second portion of a second human captured at the same time from different points of view; input the first and second pairs of images to the model; the model configured to: generate first and second reconstructed images of the at least the first portion of the first human based on the first and second pairs, respectively, and the training module is configured to selectively adjust one or more parameters of the model based on: the first reconstructed image and the second reconstructed image.

Abstract: A method includes: performing unsupervised pre-training of a model, the model including and a decoder including: obtaining a first image and a second image under different conditions or from different viewpoints; encoding, by the encoder, the first image into a representation of the first image and the second image into a representation of the second image; transforming the representation of the first image into a transformed representation; decoding, by the decoder, the transformed representation into a reconstructed image, where the transforming of the representation of the first image and the decoding of the transformed representation is based on the representation of the first image and the representation of the second image; and adjusting one or more parameters of at least one of the encoder and the decoder based on minimizing a loss; and fine-tuning the model, initialized with a set of task specific encoder parameters, for a geometric vision task.

Abstract: A system includes: a depth module including an encoder and a decoder and configured to: receive a first image from a first time from a camera; and based on the first image, generate a depth map including depths between the camera and objects in the first image; a pose module configured to: generate a first pose of the camera based on the first image; generate a second pose of the camera for a second time based on a second image; and generate a third pose of the camera for a third time based on a third image; and a motion module configured to: determine a first motion of the camera between the second and first times based on the first and second poses; and determine a second motion of the camera between the second and third times based on the second and third poses.

Abstract: A computer-implemented method of determining a position of a portable electronic device in an indoor environment includes: at a first rate, updating an absolute position of a portable electronic device within the indoor environment based on at least one of radio signal data and magnetic field data captured using the portable electronic device; at a second rate that is different than the first rate, selectively updating an estimated displacement of the portable electronic device within the indoor environment, the updating the estimated displacement comprising generating an estimated displacement, by a neural network module, based on inertial sensor data of the portable electronic device; and determining a present position of the portable electronic device within the indoor environment by updating a previous position based on at least one of (a) the estimated displacement and (b) the absolute position.

Abstract: A computer-implemented method includes: obtaining a pair of images depicting a same scene, the pair of images including a first image with a first pixel grid and a second image with a second pixel grid, the first pixel grid different than the second pixel grid; by a neural network module having a first set of parameters: generating a first feature map based on the first image; and generating a second feature map based on the second image; determining a first correlation volume based on the first and second feature maps; iteratively determining a second correlation volume based on the first correlation volume; determining a loss for the first and second feature maps based on the second correlation volume; generating a second set of the parameters based on minimizing a loss function using the loss; and updating the neural network module to include the second set of parameters.

Abstract: A method of jointly training of a transferable feature extractor network, an ordinal regressor network, and an order classifier network in an ordinal regression unsupervised domain adaption network by providing a source of labeled source images and unlabeled target images; outputting image representations from a transferable feature extractor network by performing a minimax optimization procedure on the source of labeled source images and unlabeled target images; training a domain discriminator network, using the image representations from the transferable feature extractor network, to distinguish between source images and target images; training an ordinal regressor network using a full set of source images from the transferable feature extractor network; and training an order classifier network using a full set of source images from said transferable feature extractor network.

Abstract: A computer-implemented method for training an artificial neural network with training data including samples and corresponding labels for performing a task includes: pre-training the artificial neural network to generate matrix representations that are invariant to a predetermined set of data augmentations applied to a sample, where the artificial neural network includes an encoder module and a projection module configured to generate the matrix representations based on ones of the samples, respectively; and after the pre-training, fine-tune training the artificial neural network using a loss function, wherein fine-tuning the artificial neural network includes adjusting, based on the labels, one or more weights of the projection module while maintaining constant weights of the encoder module, and where the loss function is based on a logit adjustment loss that is based on logits that are adjusted based on a class distribution of the training data.

Abstract: A pose estimation training system includes: a first model configured to generate a first 6 degrees of freedom (DoF) pose of a first camera that captured a first image from a first domain; a second model configured to generate a second 6 DoF pose of a second camera that captured a second image from a second domain, where the second domain is different than the first domain; a discriminator module configured to, based on first and second outputs from the first and second encoder modules, generate a discriminator output indicative of whether the first and second images are from the same domain; and a training control module configured to, based on the discriminator output, selectively adjust at least one weight value shared by the first model and the second model.

Abstract: A pre-trained source encoder generates a source encoder representation for each image of a labeled set of source images from a source domain. A target encoder generates a target encoder representation for each image of an unlabeled set of target images from a target domain. A generative adversarial network outputs a first prediction indicating whether each of the source encoder representations and each of the target encoder representations originate from the source domain or the target domain. The generative adversarial network outputs a second prediction of the latent code for each of the source encoder representations and each of the target encoder representations. The target encoder and the generative adversarial network are trained by repeatedly updating parameters of the target encoder and the generative adversarial network until one or more predetermined stopping conditions occur.

Abstract: A method of training a predictor to predict a location of a computing device in an indoor environment incudes: receiving training data including strength of signals received from wireless access points at positions of an indoor environment, where the training data includes: a subset of labeled data including signal strength values and location labels; and a subset of unlabeled data including signal strength values and not including labels indicative of locations; training a variational autoencoder to minimize a reconstruction loss of the signal strength values of the training data, where the variational autoencoder includes encoder neural networks and decoder neural networks; and training a classification neural network to minimize a prediction loss on the labeled data, where the classification neural network generates a predicted location based on the latent variable, and where the encoder neural networks and the classification neural network form the predictor.

Abstract: A computer-implemented method of determining a position of a portable electronic device in an indoor environment includes: at a first rate, updating an absolute position of a portable electronic device within the indoor environment based on at least one of radio signal data and magnetic field data captured using the portable electronic device; at a second rate that is different than the first rate, selectively updating an estimated displacement of the portable electronic device within the indoor environment, the updating the estimated displacement comprising generating an estimated displacement, by a neural network module, based on inertial sensor data of the portable electronic device; and determining a present position of the portable electronic device within the indoor environment by updating a previous position based on at least one of (a) the estimated displacement and (b) the absolute position.

Abstract: A computer-implemented method for generating a semantically segmented image and a depth completion image using a convolutional neural network (CNN) from an input visible image and/or an input depth image. A central component of the CNN for semantic segmentation and depth completion is a common representation that allows both tasks to be performed when given any of these combinations of input images (i) both an input visible image and an input depth image, (ii) only an input visible image, or (iii) only an input depth image.

Abstract: A system includes: a depth module including an encoder and a decoder and configured to: receive a first image from a first time from a camera; and based on the first image, generate a depth map including depths between the camera and objects in the first image; a pose module configured to: generate a first pose of the camera based on the first image; generate a second pose of the camera for a second time based on a second image; and generate a third pose of the camera for a third time based on a third image; and a motion module configured to: determine a first motion of the camera between the second and first times based on the first and second poses; and determine a second motion of the camera between the second and third times based on the second and third poses.

Abstract: A pose estimation training system includes: a first model configured to generate a first 6 degrees of freedom (DoF) pose of a first camera that captured a first image from a first domain; a second model configured to generate a second 6 DoF pose of a second camera that captured a second image from a second domain, where the second domain is different than the first domain; a discriminator module configured to, based on first and second outputs from the first and second encoder modules, generate a discriminator output indicative of whether the first and second images are from the same domain; and a training control module configured to, based on the discriminator output, selectively adjust at least one weight value shared by the first model and the second model.

Abstract: A computer-implemented method for generating a semantically segmented image and a depth completion image using a convolutional neural network (CNN) from an input visible image and/or an input depth image. A central component of the CNN for semantic segmentation and depth completion is a common representation that allows both tasks to be performed when given any of these combinations of input images (i) both an input visible image and an input depth image, (ii) only an input visible image, or (iii) only an input depth image.

Abstract: A method of training a predictor to predict a location of a computing device in an indoor environment incudes: receiving training data including strength of signals received from wireless access points at positions of an indoor environment, where the training data includes: a subset of labeled data including signal strength values and location labels; and a subset of unlabeled data including signal strength values and not including labels indicative of locations; training a variational autoencoder to minimize a reconstruction loss of the signal strength values of the training data, where the variational autoencoder includes encoder neural networks and decoder neural networks; and training a classification neural network to minimize a prediction loss on the labeled data, where the classification neural network generates a predicted location based on the latent variable, and where the encoder neural networks and the classification neural network form the predictor.

Abstract: A pre-trained source encoder generates a source encoder representation for each image of a labeled set of source images from a source domain. A target encoder generates a target encoder representation for each image of an unlabeled set of target images from a target domain. A generative adversarial network outputs a first prediction indicating whether each of the source encoder representations and each of the target encoder representations originate from the source domain or the target domain. The generative adversarial network outputs a second prediction of the latent code for each of the source encoder representations and each of the target encoder representations. The target encoder and the generative adversarial network are trained by repeatedly updating parameters of the target encoder and the generative adversarial network until one or more predetermined stopping conditions occur.

Abstract: A method and system are disclosed for dynamically estimating an origin-destination matrix. An origin-destination matrix is initialized with a set of origin stops and destination stops. Validation sequences are acquired for a set of travelers on a transportation system which include a plurality of the origin stops and respective timestamps. Corresponding destination stops may be known or inferred. For each validation sequence, a set of subsequences is generated, each including a respective one of the origin stops and the associated timestamp. Subsequences which, in combination, constitute a valid transfer trip are identified. For a combination of subsequences constituting a valid transfer trip, the method includes determining whether the valid transfer trip is a multi-goal trip for which there is least a first destination stop with an intermediate goal and a second destination stop with a final goal. The origin-destination matrix is updated, based on the determination.