Abstract: Techniques are provided in which a plurality of edges are detected within a digital image. An anchor point located along an edge of the plurality of edges is selected. An analysis grid associated with the anchor point is generated, the analysis grid including a plurality of cells. An anchor point normal vector comprising a normal vector of the edge at the anchor point is calculated. Edge pixel normal vectors comprising normal vectors of the edge at locations along the edge within the cells of the analysis grid are calculated. A histogram of similarity is generated for each of one or more cells of the analysis grid, each histogram of similarity being based on a similarity measure between each of the edge pixel normal vectors within a cell and the anchor point normal vector, and a descriptor is generated for the analysis grid based on the histograms of similarity.

Abstract: A method at a computing device for classifying elements within an input, the method including breaking the input into a plurality of patches; for each patch: creating a vector output; applying a characterization map to select a classification bin from a plurality of classification bins; and utilizing the selected classification bin to classify the vector output to create a classified output; and compiling the classified output from each patch.

Abstract: A sensor data processing system and method is described. Contemplated systems and methods derive a first recognition trait of an object from a first data set that represents the object in a first environmental state. A second recognition trait of the object is then derived from a second data set that represents the object in a second environmental state. The sensor data processing systems and methods then identifies a mapping of elements of the first and second recognition traits in a new representation space. The mapping of elements satisfies a variance criterion for corresponding elements, which allows the mapping to be used for object recognition. The sensor data processing systems and methods described herein provide new object recognition techniques that are computationally efficient and can be performed in real-time by the mobile phone technology that is currently available.

Abstract: Edge-based recognition systems and methods are presented. Edges of the object are identified from the image data based on co-circularity of edgels, and edge-based descriptors are constructed based on the identified edges. The edge-based descriptors along with additional perception metrics are used to obtain a list of candidate objects matched with the edge-based descriptors. Through various filtering processes and verification processes, false positive candidate objects are further removed from the list to determine the final candidate object.

Abstract: Systems and methods of generating a compact visual vocabulary are provided. Descriptor sets related to digital representations of objects are obtained, clustered and partitioned into cells of a descriptor space, and a representative descriptor and index are associated with each cell. Generated visual vocabularies could be stored in client-side devices and used to obtain content information related to objects of interest that are captured.

Abstract: A face image generation method includes: determining, according to a first face image, a three dimensional morphable model (3DMM) corresponding to the first face image as a first model; determining, according to a reference element, a 3DMM corresponding to the reference element as a second model, the reference element representing a posture and/or an expression of a target face image; determining, according to the first model and the second model, an initial optical flow map corresponding to the first face image, and deforming the first face image according to the initial optical flow map to obtain an initial deformation map; obtaining, through a convolutional neural network, an optical flow increment map and a visibility probability map that correspond to the first face image; and generating the target face image according to the first face image, the initial optical flow map, the optical flow increment map, and the visibility probability map.

Abstract: An example system for performing segmentation of data based on tensor inputs includes memory storing computer-executable instructions defining a learning network, where the learning network includes a plurality of sequential encoder down-sampling blocks. A processor is configured to execute the computer-executable instructions to receive a multi-dimensional input tensor including at least a first dimension, a second dimension and a plurality of channels. The processor is also configured to process the received multi-dimensional input tensor by passing the received multi-dimensional input tensor through the plurality of sequential encoder down-sampling blocks of the learning network, and to generate an output tensor in response to processing the received multi-dimensional input tensor. The output tensor includes at least one segmentation classification.

Abstract: An object recognition ingestion system is presented. The object ingestion system captures image data of objects, possibly in an uncontrolled setting. The image data is analyzed to determine if one or more a priori know canonical shape objects match the object represented in the image data. The canonical shape object also includes one or more reference PoVs indicating perspectives from which to analyze objects having the corresponding shape. An object ingestion engine combines the canonical shape object along with the image data to create a model of the object. The engine generates a desirable set of model PoVs from the reference PoVs, and then generates recognition descriptors from each of the model PoVs. The descriptors, image data, model PoVs, or other contextually relevant information are combined into key frame bundles having sufficient information to allow other computing devices to recognize the object at a later time.

Abstract: Methods, systems, and articles of manufacture to improve image recognition searching are disclosed. In some embodiments, a first document image of a known object is used to generate one or more other document images of the same object by applying one or more techniques for synthetically generating images. The synthetically generated images correspond to different variations in conditions under which a potential query image might be captured. Extracted features from an initial image of a known object and features extracted from the one or more synthetically generated images are stored, along with their locations, as part of a common model of the known object. In other embodiments, image recognition search effectiveness is improved by transforming the location of features of multiple images of a same known object into a common coordinate system. This can enhance the accuracy of certain aspects of existing image search/recognition techniques including, for example, geometric verification.

Abstract: Techniques are provided for determining a cell count within a whole slide pathology image. The image is segmented using a global threshold value to define a tissue area. A plurality of patches comprising the tissue area are selected. Stain intensity vectors are determined within the plurality of patches to generate a stain intensity image. The stain intensity image is iteratively segmented to generate a cell mask using a local threshold value that is and gradually reduced after each iteration. A chamfer distance transform is applied to the cell mask to generate a distance map. Cell seeds are determined on the distance map. Cell segments are determined using a watershed transformation, and a whole cell count is calculated for the plurality of patches based on the cell segments. A client device may be configured for real-time cell counting based on the whole cell count.

Abstract: Methods, systems, and articles of manufacture to improve image recognition searching are disclosed. In some embodiments, a first document image of a known object is used to generate one or more other document images of the same object by applying one or more techniques for synthetically generating images. The synthetically generated images correspond to different variations in conditions under which a potential query image might be captured. Extracted features from an initial image of a known object and features extracted from the one or more synthetically generated images are stored, along with their locations, as part of a common model of the known object. In other embodiments, image recognition search effectiveness is improved by transforming the location of features of multiple images of a same known object into a common coordinate system. This can enhance the accuracy of certain aspects of existing image search/recognition techniques including, for example, geometric verification.

Abstract: An object recognition ingestion system is presented. The object ingestion system captures image data of objects, possibly in an uncontrolled setting. The image data is analyzed to determine if one or more a priori know canonical shape objects match the object represented in the image data. The canonical shape object also includes one or more reference PoVs indicating perspectives from which to analyze objects having the corresponding shape. An object ingestion engine combines the canonical shape object along with the image data to create a model of the object. The engine generates a desirable set of model PoVs from the reference PoVs, and then generates recognition descriptors from each of the model PoVs. The descriptors, image data, model PoVs, or other contextually relevant information are combined into key frame bundles having sufficient information to allow other computing devices to recognize the object at a later time.

Abstract: Techniques are provided that include receiving one or more global signatures for a query image in response to an image recognition query, wherein some of the plurality of global signatures are generated using local descriptors corresponding to different cropped versions of the image. A ranking order is determined for a plurality of document images based on nearest neighbor relations between document signatures corresponding to the plurality of document images and each one of the one or more global signatures for the query image. A subset of the plurality of document images is selected based on the determined ranking order. Additional document data corresponding to the selected subset of the plurality of document images is obtained, and a search result is generated based on a geometric verification between the additional document data corresponding to the selected subset of the plurality of document images and the query image.

Abstract: Multiparty object recognition systems and methods are disclosed. A method of interactively manipulating virtual object data, wherein an object database is configured to store first party object data that corresponds to a first real-world object and is further configured to store second party object data that corresponds to a second real-world object, includes obtaining the first party object data and the second party object data for storage within the object database. Access to the object database is controlled such that the first party object data and the second party object data is accessible to the first party and the second party. Modification of the first party object data by the second party is facilitated to generate modified first party object data that is in accordance with at least one context parameter of the second party object data, and the modified first party object data is communicated to the first party.

Abstract: Techniques are provided in which a plurality of edges are detected within a digital image. An anchor point located along an edge of the plurality of edges is selected. An analysis grid associated with the anchor point is generated, the analysis grid including a plurality of cells. An anchor point normal vector comprising a normal vector of the edge at the anchor point is calculated. Edge pixel normal vectors comprising normal vectors of the edge at locations along the edge within the cells of the analysis grid are calculated. A histogram of similarity is generated for each of one or more cells of the analysis grid, each histogram of similarity being based on a similarity measure between each of the edge pixel normal vectors within a cell and the anchor point normal vector, and a descriptor is generated for the analysis grid based on the histograms of similarity.

Abstract: A system capable of determining which recognition algorithms should be applied to regions of interest within digital representations is presented. A preprocessing module utilizes one or more feature identification algorithms to determine regions of interest based on feature density. The preprocessing modules leverages the feature density signature for each region to determine which of a plurality of diverse recognition modules should operate on the region of interest. A specific embodiment that focuses on structured documents is also presented. Further, the disclosed approach can be enhanced by addition of an object classifier that classifies types of objects found in the regions of interest.

Abstract: A computer implemented method of generating at least one shape of a region of interest in a digital image is provided.

Abstract: Apparatus, methods and systems of providing AR content are disclosed. Embodiments of the inventive subject matter can obtain an initial map of an area, derive views of interest, obtain AR content objects associated with the views of interest, establish experience clusters and generate a tile map tessellated based on the experience clusters. A user device could be configured to obtain and instantiate at least some of the AR content objects based on at least one of a location and a recognition.

Abstract: Systems and methods of generating a compact visual vocabulary are provided. Descriptor sets related to digital representations of objects are obtained, clustered and partitioned into cells of a descriptor space, and a representative descriptor and index are associated with each cell. Generated visual vocabularies could be stored in client-side devices and used to obtain content information related to objects of interest that are captured.

Abstract: Multiparty object recognition systems and methods are disclosed. A method of interactively manipulating virtual object data, wherein an object database is configured to store first party object data that corresponds to a first real-world object and is further configured to store second party object data that corresponds to a second real-world object, includes obtaining the first party object data and the second party object data for storage within the object database. Access to the object database is controlled such that the first party object data and the second party object data is accessible to the first party and the second party. Modification of the first party object data by the second party is facilitated to generate modified first party object data that is in accordance with at least one context parameter of the second party object data, and the modified first party object data is communicated to the first party.