A computer implemented image processing method is provided for identifying a tissue boundary of a tumor region of a tissue sample, the tissue sample containing non-tumor regions and at least one tumor region. The method includes obtaining an image of a tissue section of the tissue sample, identifying at least one image property of the image, and comparing the image property with classification data. The method further includes, based on the comparison, classifying a region of the image as a tumor region representing a tumor region in the tissue sample or a non-tumor region representing a non-tumor region in the tissue sample. If the region of the image is classified as a tumor region, the method includes identifying a boundary of the region of the image, and using the boundary to identify a tissue boundary of the tumor region of the tissue sample represented by the region of the image.

A method for counting individuals in an image containing a dense, uniform or non-uniform crowd. The current invention leverages multiple sources of information to compute an estimate of the number of individuals present in a dense crowd visible in a single image. This approach relies on multiple sources, such as low confidence head detections, repetition of texture elements (using SIFT), and frequency-domain analysis to estimate counts, along with confidence associated with observing individuals in an image region. Additionally, a global consistency constraint can be employed on counts using Markov Random Field. This caters for disparity in counts in local neighborhoods and across scales. The methodology was tested on a new dataset of fifty (50) crowd images containing over 64,000 annotated humans, with the head counts ranging from 94 to 4,543. Efficient and accurate results were attained.

A process and apparatus are provided to characterize low-resolution partial point clouds for object recognition or query. A partial point cloud representation of an object is received. Zero and first order geometric moments of the partial point cloud are computed. A location of a center of a point cloud mass is computed using the geometric moments. A cubic bounding box is generated centered at the location of the mass center of the point cloud, with one side of the box bounding the point cloud at its longest semi-axis. The bounding box is divided into a three dimensional grid. A normalized voxel mass distribution is generated over the three dimensional grid. Tchebichef moments of different orders are calculated with respect to the voxel mass distribution in the grid. Low-order moments are collected to form TMSDs. Similarity is compared between the TMSD of the point cloud with TMSDs of other point clouds.

A method and an image processing apparatus for image-based object feature description are provided. In the method, an object of interest in an input image is detected and a centroid and a direction angle of the object of interest are calculated. Next, a contour of the object of interest is recognized and a distance and a relative angle of each pixel on the contour to the centroid are calculated, in which the relative angle of each pixel is calibrated by using the direction angle. Then, a 360-degree range centered on the centroid is equally divided into multiple angle intervals and the pixels on the contour are separated into multiple groups according to a range covered by each angle interval. Afterwards, a maximum among the distances of the pixels in each group is obtained and used as a feature value of the group. Finally, the feature values of the groups are normalized and collected to form a feature vector that serves as a feature descriptor of the object of interest.

A method for counting individuals in an image containing a dense, uniform or non-uniform crowd. The current invention leverages multiple sources of information to compute an estimate of the number of individuals present in a dense crowd visible in a single image. This approach relies on multiple sources, such as low confidence head detections, repetition of texture elements (using SIFT), and frequency-domain analysis to estimate counts, along with confidence associated with observing individuals in an image region. Additionally, a global consistency constraint can be employed on counts using Markov Random Field. This caters for disparity in counts in local neighborhoods and across scales. The methodology was tested on a new dataset of fifty (50) crowd images containing over 64,000 annotated humans, with the head counts ranging from 94 to 4,543. Efficient and accurate results were attained.

Aspects of the disclosure relate to user authentication. A computing platform may receive a plurality of facial scans of an individual. The computing platform may train, using the plurality of facial scans, a convolutional neural network (CNN) to identify the individual, based on a first facial scan of the individual, using subclasses of the CNN. The computing platform may receive an authorization request including the first facial scan of the individual. The computing platform may input the first facial scan into the CNN, which may cause the CNN to identify the individual. Based on successful identification of the individual, the computing platform may grant requested access to the individual. The computing platform may update, using the first facial scan, the CNN.

Aspects of the disclosure relate to user authentication. A computing platform may receive a plurality of facial scans of an individual. The computing platform may train, using the plurality of facial scans, a convolutional neural network (CNN) to identify the individual, based on a first facial scan of the individual, using subclasses of the CNN. The computing platform may receive an authorization request including the first facial scan of the individual. The computing platform may input the first facial scan into the CNN, which may cause the CNN to identify the individual. Based on successful identification of the individual, the computing platform may grant requested access to the individual. The computing platform may update, using the first facial scan, the CNN.

A system for performing global-scale damage detection using satellite imagery, comprising a damage detection server that receives and analyzes image data to identify objects within an image via a curated computational method, and a curation interface that enables a user to curate image information for use in object identification, and a method for a curated computational method for performing global scale damage detection.

An architecture for linear-time extraction of maximally stable extremal regions (MSERs) having an image memory, heap memory, a pointer array and processing hardware is disclosed. The processing hardware is configured to in real-time analyze image pixels in the image memory using a linear-time algorithm to identify a plurality of components of the image. The processing hardware is also configured to place the image pixels in the heap memory for each of the plurality of components of the image, generate a pointer that points to a location in the heap memory that is associated with a start of flooding for another component and store the pointer in the array of pointers. The processing hardware is also configured to access the plurality of components using the array of pointers and determine MSER ellipses based on the components and MSER criteria.

A method for identifying bleeding in a patient, comprising: receiving a four-dimensional data set comprising vascular voxels, each of the voxels representing a three-dimensional location of a time-varying signal; identifying the vascular voxels in the data set by comparing each time-varying signal of each voxel in the data set to a variance threshold; identifying, within the vascular voxels in the data set, blood vessel site voxels by comparing each time-varying signal of the vascular voxels to a model arterial and venous signals; generating clusters of voxels in the data set; separating the clusters of voxels into a subset of blood vessel site clusters and a subset of remaining clusters; identifying, within the subset of remaining clusters, one or more clusters that are spatially growing over time to determine one or more active bleed sites in the patient; and generating an alert if one or more active bleed sites are determined.