Abstract: Classification algorithm optimization is described. In an example, a classification algorithm is optimized by calculating an evaluation sequence for a set of weighted feature functions that orders the feature functions in accordance with a measure of influence on the classification algorithm. Classification thresholds are determined for each step of the evaluation sequence, which indicate whether a classification decision can be made early and the classification algorithm terminated without evaluating further feature functions. In another example, a classifier applies the weighted feature functions to previously unseen data in the order of the evaluation sequence and determines a cumulative value at each step. The cumulative value is compared to the classification thresholds at each step to determine whether a classification decision can be made early without evaluating further feature functions.

Abstract: Techniques for human body pose estimation are disclosed herein. Images such as depth images, silhouette images, or volumetric images may be generated and pixels or voxels of the images may be identified. The techniques may process the pixels or voxels to determine a probability that each pixel or voxel is associated with a segment of a body captured in the image or to determine a three-dimensional representation for each pixel or voxel that is associated with a location on a canonical body. These probabilities or three-dimensional representations may then be utilized along with the images to construct a posed model of the body captured in the image.

Abstract: Checking and/or completing for data grids is described such as for grids having rows and columns of cells at least some of which contain data values such as numbers or categories. In various embodiments predictive probability distributions are obtained from an inference engine for one or more of the cells and the predictive probability distributions are used for various tasks such as to suggest values to complete blank cells, highlight cells having outlying values, identify potential errors, suggest corrections to potential errors, identify similarities between cells, identify differences between cells, cluster rows of the data grid, and other tasks. In various embodiments a graphical user interface displays a data grid and provides facilities for completing, error checking/correcting, and analyzing data in the data grid.

Abstract: Image completion using scene geometry is described, for example, to remove marks from digital photographs or complete regions which are blank due to editing. In an embodiment an image depicting, from a viewpoint, a scene of textured objects has regions to be completed. In an example, geometry of the scene is estimated from a depth map and the geometry used to warp the image so that at least some surfaces depicted in the image are fronto-parallel to the viewpoint. An image completion process is guided using distortion applied during the warping. For example, patches used to fill the regions are selected on the basis of distortion introduced by the warping. In examples where the scene comprises regions having only planar surfaces the warping process comprises rotating the image. Where the scene comprises non-planar surfaces, geodesic distances between image elements may be scaled to flatten the non-planar surfaces.

Abstract: Three-dimensional environment reconstruction is described. In an example, a 3D model of a real-world environment is generated in a 3D volume made up of voxels stored on a memory device. The model is built from data describing a camera location and orientation, and a depth image with pixels indicating a distance from the camera to a point in the environment. A separate execution thread is assigned to each voxel in a plane of the volume. Each thread uses the camera location and orientation to determine a corresponding depth image location for its associated voxel, determines a factor relating to the distance between the associated voxel and the point in the environment at the corresponding location, and updates a stored value at the associated voxel using the factor. Each thread iterates through an equivalent voxel in the remaining planes of the volume, repeating the process to update the stored value.

Abstract: Use of a 3D environment model in gameplay is described. In an embodiment, a mobile depth camera is used to capture a series of depth images as it is moved around and a dense 3D model of the environment is generated from this series of depth images. This dense 3D model is incorporated within an interactive application, such as a game. The mobile depth camera is then placed in a static position for an interactive phase, which in some examples is gameplay, and the system detects motion of a user within a part of the environment from a second series of depth images captured by the camera. This motion provides a user input to the interactive application, such as a game. In further embodiments, automatic recognition and identification of objects within the 3D model may be performed and these identified objects then change the way that the interactive application operates.

Abstract: Predicting joint positions is described, for example, to find joint positions of humans or animals (or parts thereof) in an image to control a computer game or for other applications. In an embodiment image elements of a depth image make joint position votes so that for example, an image element depicting part of a torso may vote for a position of a neck joint, a left knee joint and a right knee joint. A random decision forest may be trained to enable image elements to vote for the positions of one or more joints and the training process may use training images of bodies with specified joint positions. In an example a joint position vote is expressed as a vector representing a distance and a direction of a joint position from an image element making the vote. The random decision forest may be trained using a mixture of objectives.

Abstract: Real-time camera tracking using depth maps is described. In an embodiment depth map frames are captured by a mobile depth camera at over 20 frames per second and used to dynamically update in real-time a set of registration parameters which specify how the mobile depth camera has moved. In examples the real-time camera tracking output is used for computer game applications and robotics. In an example, an iterative closest point process is used with projective data association and a point-to-plane error metric in order to compute the updated registration parameters. In an example, a graphics processing unit (GPU) implementation is used to optimize the error metric in real-time. In some embodiments, a dense 3D model of the mobile camera environment is used.

Abstract: A tractable model solves certain labeling problems by providing potential functions having arbitrary dependencies upon an observed dataset (e.g., image data). The model uses decision trees corresponding to various factors to map dataset content to a set of parameters used to define the potential functions in the model. Some factors define relationships among multiple variable nodes. When making label predictions on a new dataset, the leaf nodes of the decision tree determine the effective weightings for such potential functions. In this manner, decision trees define non-parametric dependencies and can represent rich, arbitrary functional relationships if sufficient training data is available. Decision trees training is scalable, both in the training set size and by parallelization. Maximum pseudolikelihood learning can provide for joint training of aspects of the model, including feature test selection and ordering, factor weights, and the scope of the interacting variable nodes used in the graph.

Abstract: Learning image processing tasks from scene reconstructions is described where the tasks may include but are not limited to: image de-noising, image in-painting, optical flow detection, interest point detection. In various embodiments training data is generated from a 2 or higher dimensional reconstruction of a scene and from empirical images of the same scene. In an example a machine learning system learns at least one parameter of a function for performing the image processing task by using the training data. In an example, the machine learning system comprises a random decision forest. In an example, the scene reconstruction is obtained by moving an image capture apparatus in an environment where the image capture apparatus has an associated dense reconstruction and camera tracking system.

Abstract: Using high-level attributes to guide image processing is described. In an embodiment high-level attributes of images of people such as height, torso orientation, body shape, gender are used to guide processing of the images for various tasks including but not limited to joint position detection, body part classification, medical image analysis and others. In various embodiments one or more random decision forests are trained using images where global variable values such as player height are known in addition to ground-truth data appropriate for the image processing task concerned. In some examples sequences of images are used where global variables are static or vary smoothly over the sequence. In some examples one or more trained random decision forests are used to find global variable values as well as output values for the task concerned such as joint positions or body part classes.

Abstract: Surface segmentation from RGB and depth images is described. In one example, a computer receives an image of a scene. The image has pixels which each have an associated color value and an associated depth value representing a distance between from an image sensor to a surface in the scene. The computer uses the depth values to derive a set of three-dimensional planes present within the scene. A cost function is used to determine whether each pixel belongs to one of the planes, and the image elements are labeled accordingly. The cost function has terms dependent on the depth value of a pixel, and the color values of the pixels and at least one neighboring pixel. In various examples, the planes can be extended until they intersect to determine the extent of the scene, and pixels not belonging to a plane can be labeled as objects on the surfaces.

Abstract: Real-time camera tracking using depth maps is described. In an embodiment depth map frames are captured by a mobile depth camera at over 20 frames per second and used to dynamically update in real-time a set of registration parameters which specify how the mobile depth camera has moved. In examples the real-time camera tracking output is used for computer game applications and robotics. In an example, an iterative closest point process is used with projective data association and a point-to-plane error metric in order to compute the updated registration parameters. In an example, a graphics processing unit (GPU) implementation is used to optimize the error metric in real-time. In some embodiments, a dense 3D model of the mobile camera environment is used.

Abstract: Moving object segmentation using depth images is described. In an example, a moving object is segmented from the background of a depth image of a scene received from a mobile depth camera. A previous depth image of the scene is retrieved, and compared to the current depth image using an iterative closest point algorithm. The iterative closest point algorithm includes a determination of a set of points that correspond between the current depth image and the previous depth image. During the determination of the set of points, one or more outlying points are detected that do not correspond between the two depth images, and the image elements at these outlying points are labeled as belonging to the moving object. In examples, the iterative closest point algorithm is executed as part of an algorithm for tracking the mobile depth camera, and hence the segmentation does not add substantial additional computational complexity.

Abstract: In one or more implementations, a static geometry model is generated, from one or more images of a physical environment captured using a camera, using one or more static objects to model corresponding one or more objects in the physical environment. Interaction of a dynamic object with at least one of the static objects is identified by analyzing at least one image and a gesture is recognized from the identified interaction of the dynamic object with the at least one of the static objects to initiate an operation of the computing device.

Abstract: Generating computer models of 3D objects is described. In one example, depth images of an object captured by a substantially static depth camera are used to generate the model, which is stored in a memory device in a three-dimensional volume. Portions of the depth image determined to relate to the background are removed to leave a foreground depth image. The position and orientation of the object in the foreground depth image is tracked by comparison to a preceding depth image, and the foreground depth image is integrated into the volume by using the position and orientation to determine where to add data derived from the foreground depth image into the volume. In examples, the object is hand-rotated by a user before the depth camera. Hands that occlude the object are integrated out of the model as they do not move in sync with the object due to re-gripping.

Abstract: An enhanced training sample set containing new synthesized training images that are artificially generated from an original training sample set is provided to satisfactorily increase the accuracy of an object recognition system. The original sample set is artificially augmented by introducing one or more variations to the original images with little to no human input. There are a large number of possible variations that can be introduced to the original images, such as varying the image's position, orientation, and/or appearance and varying an object's context, scale, and/or rotation. Because there are computational constraints on the amount of training samples that can be processed by object recognition systems, one or more variations that will lead to a satisfactory increase in the accuracy of the object recognition performance are identified and introduced to the original images.

Abstract: Image completion using scene geometry is described, for example, to remove marks from digital photographs or complete regions which are blank due to editing. In an embodiment an image depicting, from a viewpoint, a scene of textured objects has regions to be completed. In an example, geometry of the scene is estimated from a depth map and the geometry used to warp the image so that at least some surfaces depicted in the image are fronto-parallel to the viewpoint. An image completion process is guided using distortion applied during the warping. For example, patches used to fill the regions are selected on the basis of distortion introduced by the warping. In examples where the scene comprises regions having only planar surfaces the warping process comprises rotating the image. Where the scene comprises non-planar surfaces, geodesic distances between image elements may be scaled to flatten the non-planar surfaces.

Abstract: Classification algorithm optimization is described. In an example, a classification algorithm is optimized by calculating an evaluation sequence for a set of weighted feature functions that orders the feature functions in accordance with a measure of influence on the classification algorithm. Classification thresholds are determined for each step of the evaluation sequence, which indicate whether a classification decision can be made early and the classification algorithm terminated without evaluating further feature functions. In another example, a classifier applies the weighted feature functions to previously unseen data in the order of the evaluation sequence and determines a cumulative value at each step. The cumulative value is compared to the classification thresholds at each step to determine whether a classification decision can be made early without evaluating further feature functions.

Abstract: Predicting joint positions is described, for example, to find joint positions of humans or animals (or parts thereof) in an image to control a computer game or for other applications. In an embodiment image elements of a depth image make joint position votes so that for example, an image element depicting part of a torso may vote for a position of a neck joint, a left knee joint and a right knee joint. A random decision forest may be trained to enable image elements to vote for the positions of one or more joints and the training process may use training images of bodies with specified joint positions. In an example a joint position vote is expressed as a vector representing a distance and a direction of a joint position from an image element making the vote. The random decision forest may be trained using a mixture of objectives.