Abstract: Techniques are provided for efficiently generating 3D information from a set of digital images. Techniques are also provided for displaying groups (or clusters) of digital images using 3D information associated with the digital images. In one technique, a group of digital images are displayed as a stack of thumbnail images where the thumbnail images are aligned on a display with respect to each other based on common features identified in the digital images, camera position, and/or camera pose. In another technique, a group of digital images are organized on a display in either a 3D layout or a 2D layout based on 3D information associated with each digital image in the group. In another technique, a transition effect is generated based on projections of two digital images onto a common scene plane and blending (or cross fading) one of the 3D projections with the other of the 3D projections.

Abstract: Systems, methods, and computer readable media for adjusting the orientation of an image frame and a scene depicted in the image frame are described. In general, techniques are disclosed for analyzing an image with one or more feature detectors to identify features in the image. An alignment or position associated with one or more features identified in the image may be used to determine a proper orientation for the image frame. The image can then be rotated to the proper orientation. It may also be determined if a scene depicted in the image is properly aligned in the rotated image orientation. If not, alignment information associated with the identified features may be utilized to straighten the depicted scene.

Abstract: Techniques are provided for efficiently generating 3D information from a set of digital images. Techniques are also provided for displaying groups (or clusters) of digital images using 3D information associated with the digital images. In one technique, a group of digital images are displayed as a stack of thumbnail images where the thumbnail images are aligned on a display with respect to each other based on common features identified in the digital images, camera position, and/or camera pose. In another technique, a group of digital images are organized on a display in either a 3D layout or a 2D layout based on 3D information associated with each digital image in the group. In another technique, a transition effect is generated based on projections of two digital images onto a common scene plane and blending (or cross fading) one of the 3D projections with the other of the 3D projections.

Abstract: Systems, methods, and computer readable media for adjusting the orientation of an image frame and a scene depicted in the image frame are described. In general, techniques are disclosed for analyzing an image with one or more feature detectors to identify features in the image. An alignment or position associated with one or more features identified in the image may be used to determine a proper orientation for the image frame. The image can then be rotated to the proper orientation. It may also be determined if a scene depicted in the image is properly aligned in the rotated image orientation. If not, alignment information associated with the identified features may be utilized to straighten the depicted scene.

Abstract: Techniques are provided to improve the performance and accuracy of landmark point detection using a Constrained Local Model. The accuracy of feature filters used by the model may be improved by supplying positive and negative sets of image data from training image regions of varying shapes and sizes to a linear support vector machine training algorithm. The size and shape of regions within which a feature filter is to be applied may be determined based on a variance in training image data for a landmark point with which the feature filter is associated. A sample image may be normalized and a confidence map generated for each landmark point by applying the feature filters as a convolution on the normalized image. A vector flow map may be pre-computed to improve the efficiency with which a mean landmark point is adjusted toward a corresponding landmark point in a sample image.

Abstract: Systems, methods, and computer readable media for adjusting the orientation of an image frame and a scene depicted in the image frame are described. In general, techniques are disclosed for analyzing an image with one or more feature detectors to identify features in the image. An alignment or position associated with one or more features identified in the image may be used to determine a proper orientation for the image frame. The image can then be rotated to the proper orientation. It may also be determined if a scene depicted in the image is properly aligned in the rotated image orientation. If not, alignment information associated with the identified features may be utilized to straighten the depicted scene.

Abstract: A method, device, system, and computer program for object recognition of a 3D object of a certain object class using a statistical shape model for recovering 3D shapes from a 2D representation of the 3D object and comparing the recovered 3D shape with known 3D to 2D representations of at least one object of the object class.

Abstract: Techniques are provided to improve the performance and accuracy of landmark point detection using a Constrained Local Model. The accuracy of feature filters used by the model may be improved by supplying positive and negative sets of image data from training image regions of varying shapes and sizes to a linear support vector machine training algorithm. The size and shape of regions within which a feature filter is to be applied may be determined based on a variance in training image data for a landmark point with which the feature filter is associated. A sample image may be normalized and a confidence map generated for each landmark point by applying the feature filters as a convolution on the normalized image. A vector flow map may be pre-computed to improve the efficiency with which a mean landmark point is adjusted toward a corresponding landmark point in a sample image.

Abstract: Techniques are provided for efficiently generating 3D information from a set of digital images. Techniques are also provided for displaying groups (or clusters) of digital images using 3D information associated with the digital images. In one technique, a group of digital images are displayed as a stack of thumbnail images where the thumbnail images are aligned on a display with respect to each other based on common features identified in the digital images, camera position, and/or camera pose. In another technique, a group of digital images are organized on a display in either a 3D layout or a 2D layout based on 3D information associated with each digital image in the group. In another technique, a transition effect is generated based on projections of two digital images onto a common scene plane and blending (or cross fading) one of the 3D projections with the other of the 3D projections.

Abstract: Techniques for automatically identifying famous people and/or iconic images are provided. Object descriptors (or “faceprints”) of the famous people and iconic images are generated and “shipped” with a digital image management application used by an end-user. The digital image management application analyzes a digital image (generated, for example, by a digital camera operated by the end-user) to detect an object, such as a face, and generates a faceprint. The digital image management application compares the faceprint to the faceprints of the famous people and/or iconic images. If a match is found, then data that identifies the corresponding person or object is displayed to the end-user.

Abstract: Systems, methods, and computer readable media for adjusting the orientation of an image frame and a scene depicted in the image frame are described. In general, techniques are disclosed for analyzing an image with one or more feature detectors to identify features in the image. An alignment or position associated with one or more features identified in the image may be used to determine a proper orientation for the image frame. The image can then be rotated to the proper orientation. It may also be determined if a scene depicted in the image is properly aligned in the rotated image orientation. If not, alignment information associated with the identified features may be utilized to straighten the depicted scene.

Abstract: A technique for combining multiple individual feature detectors to identify a combined feature in a digital image is disclosed. A combined feature detection rule may specify multiple individual feature detectors with which an image is to be analyzed. The multiple individual feature detectors may identify constituent parts of the combined feature and/or may identify features based on different image properties. An analysis of the image with the specified feature detectors may result in the identification of multiple candidate regions (i.e., regions within which the detectors identify their respective features). The combined feature detection rule may operate directly on the multiple candidate regions to adjust the spatial properties of the candidate regions and group the adjusted candidate regions into candidate region groups, it may then be determined if one or more of the candidate region groups is representative of a presence of the combined feature in the image.

Abstract: The electronic device with one or more processors and memory provides a digital photograph of a real-world scene. The electronic device provides a natural language text string corresponding to a speech input associated with the digital photograph. The electronic device performs natural language processing on the text string to identify one or more terms associated with an entity, an activity, or a location. The electronic device tags the digital photograph with the one or more terms and their associated entity, activity, or location.

Abstract: Methods and apparatus for a roof analysis tool for constructing a parameter set, where the parameter set is derived from mapping data for a map region, and where the parameter set describes the roofs for the buildings within the map region. In some cases, the parameter set includes a list of roof type identification values and the respective buildings in the map region for which a given roof type identification value corresponds. The roof analysis tool may operate on a server and work in conjunction with a mobile device, where the mobile device may display map views of a map region such that the map view is based on a three-dimensional model of the map region, and where a portion of the three-dimensional model is based on data generated on the mobile device and a portion of the three-dimensional model is based on data generated on the server.

Abstract: Systems, methods, and computer readable media for determining and applying face recognition parameter sets are described. In general, techniques are disclosed for identifying and constructing a unique combination of facial recognition discriminators into a “face feature vector” that has been found to be more robust (e.g., stable to image noise, a person's pose, and scene illumination) and accurate (e.g., provide high recognition rates) than prior art techniques. More particularly, a face feature vector may be generated by the combination of shape descriptors (e.g., as generated by two-dimensional and three-dimensional shape models) and texture descriptors (e.g., as generated by global and local texture models).

Abstract: By providing 3D representations of noteworthy locations for comparison with images, the 3D location of the imaging device, as well as the orientation of the device may be determined. The 3D location and orientation of the imaging device then allows for enhanced navigation in a collection of images, as well as enhanced visualization and editing capabilities. The 3D representations of noteworthy locations may be provided in a database that may be stored local or remote to the imaging device or a programmable device processing images obtained from the imaging device.

Abstract: A method of computing global-to-local metrics for recognition. Based on training examples with feature representations, the method automatically computes a local metric that varies over the space of feature representations to optimize discrimination and the performance of recognition systems. Given a set of points in an arbitrary features space, local metrics are learned in a hierarchical manner that give low distances between points of same class and high distances between points of different classes. Rather than considering a global metric, a class-based metric or a point-based metric, the proposed invention applies successive clustering to the data and associates a metric to each one of the clusters.

Abstract: Systems, methods, and computer readable media for determining and applying face recognition parameter sets are described. In general, techniques are disclosed for identifying and constructing a unique combination of facial recognition discriminators into a “face feature vector” that has been found to be more robust (e.g., stable to image noise, a person's pose, and scene illumination) and accurate (e.g., provide high recognition rates) than prior art techniques. More particularly, a face feature vector may be generated by the combination of shape descriptors (e.g., as generated by two-dimensional and three-dimensional shape models) and texture descriptors (e.g., as generated by global and local texture models).

Abstract: Techniques for automatically identifying famous people and/or iconic images are provided. Object descriptors (or “faceprints”) of the famous people and iconic images are generated and “shipped” with a digital image management application used by an end-user. The digital image management application analyzes a digital image (generated, for example, by a digital camera operated by the end-user) to detect an object, such as a face, and generates a faceprint. The digital image management application compares the faceprint to the faceprints of the famous people and/or iconic images. If a match is found, then data that identifies the corresponding person or object is displayed to the end-user.

Abstract: One embodiment may take the form of a method of operating a computing device in a reduced power state and collecting a first set of data from at least one sensor. Based on the first set of data, the computing device determines a probability that an object is within a threshold distance of the computing device and, if so, the device activates at least one secondary sensor to collect a second set of data. Based on the second set of data, the device determines if the object is a person. If it is a person, a position of the person relative to the computing device is determined and the computing device changes its state based on the position of the person. If the object is not a person, the computing device remains in a reduced power state.