Embodiments relate to generating keypoint descriptors of the keypoints. An apparatus includes a pyramid image generator circuit and a keypoint descriptor generator circuit. The pyramid image generator circuit generates an image pyramid from an input image. The keypoint descriptor generator circuit determines intensity values of sample points in the pyramid images for a keypoint and determines comparison results of comparisons between the intensity values of pairs of the sample points. The keypoint descriptor generator circuit generate bit values defining the comparison results for the keypoint, each bit value corresponding with one of the comparison results, and generate a sequence of the bit values defining an ordering of the comparison results based on importance levels of the comparisons, where the importance level of each comparison defines how much the comparison is representative of features. Bit values for comparisons having the lowest importance levels may be excluded from the sequence.

An object detection network includes: a hybrid voxel feature extractor configured to acquire a raw point cloud, extract a hybrid scale voxel feature from the raw point cloud, and project the hybrid scale voxel feature to generate a pseudo-image feature map; a backbone network configured to perform a hybrid voxel scale feature fusion by using the pseudo-image feature map to generate multi-class pyramid features; and a detection head configured to predict a three-dimensional object box of a corresponding class according to the multi-class pyramid features. The object detection network can effectively solve a problem that under a single voxel scale, inference time is longer if the voxel scale is smaller, and an intricate feature cannot be captured and a smaller object cannot be accurately located if the voxel scale is larger. Different classes of 3D objects can be detected quickly and accurately in a 3D scene.

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for connected contact tracing system. The methods, systems, and apparatus include actions of receiving sensor data from multiple sensors, receiving exposure information including a person and an epidemiological event data, determining, from the sensor data, a contact exposure event including the person and another person, generating, from the sensor data and the exposure information, a risk score for the contact exposure event, and providing a notification including the risk score and information for the contact exposure event.

Efficient processing of chroma-subsampled video is performed using convolutional neural networks (CNNs) in which the luma and chroma channels are processed separately. The luma channel is independently convolved and downsampled and, in parallel, the chroma channels are convolved and then merged with the downsampled luma to generate encoded chroma-subsampled video. Further processing of the encoded video that involves deconvolution and upsampling, splitting into two sets of channels, and further deconvolutions and upsampling is used in CNNs to generate decoded chroma-subsampled video in compression-decompression applications, to remove noise from chroma-subsampled video, or to upsample chroma-subsampled video to RGB 444 video. CNNs with separate luma and chroma processing in which the further processing includes additional convolutions and downsampling may be used for object recognition and semantic search in chroma-subsampled video.

Embodiments relate to generating keypoint descriptors of the keypoints using a sub-scale refinement and a sample pattern radius adjustment. An apparatus includes a sub-pixel refiner circuit and a keypoint descriptor generator circuit. The sub-pixel refiner circuit determines a keypoint scale value for a scale dimension of a keypoint in an image pyramid by performing an interpolation of response map (RM) pixel values of a pixel block of RM images defined around the keypoint. The keypoint descriptor generator circuit determines sample scales of the image pyramid based on the keypoint scale value and determines a radius value for each sample scale based on the keypoint scale value. The keypoint descriptor generator circuit samples patches of pixel values at the sample scales using the radius value for each sample scale to generate a keypoint descriptor of the keypoint.

Generating magnetic resonance (MR) images of a subject from MR data obtained by a magnetic resonance imaging (MRI) system by: generating first and second sets of one or more MR images from first and second input MR data; aligning the first and second sets of MR images using a neural network model comprising first and second neural networks, the aligning comprising: estimating, using the first neural network, a first transformation between the first and second sets of MR images; generating a first updated set of MR images from the second set of MR images using the first transformation; estimating, using the second neural network, a second transformation between the first set and the first updated set of MR images; and aligning the first set of MR images and the second set of MR images at least in part by using the first transformation and the second transformation.

A method includes: determining at least one typical object ratio from a first training data set by counting ratios of objects in training pictures of the first training data set; determining at least one picture scaling size based at least on the at least one typical object ratio; scaling the training pictures of the first training data set according to the at least one picture scaling size; obtaining a second training data set by slicing the scaled training pictures; training an object detection model using the second training data set; and performing object detection on a to-be-detected picture using the trained object detection model. The object detection method according to the embodiments of the present disclosure can be used to complete, without manual intervention, a task of detecting an extremely small object.

Described are systems and methods for training a machine-learning model to generate image of biological samples, and systems and methods for generating enhanced images of biological samples. The method for training a machine-learning model to generate images of biological samples may include obtaining a plurality of training images comprising a training image of a first type, and a training image of a second type. The method may also include generating, based on the training image of the first type, a plurality of wavelet coefficients using the machine-learning model; generating, based on the plurality of wavelet coefficients, a synthetic image of the second type; comparing the synthetic image of the second type with the training image of the second type; and updating the machine-learning model based on the comparison.

A similar damage search device includes a database that stores first damage information generated on the basis of a damage image of a structure, the first damage information including a damage vector obtained by vectorizing damage of the structure, and damage structure information including at least one of information on a hierarchical structure of the damage vector or information on a direction of the damage vector, an information acquisition unit that acquires second damage information corresponding to the first damage information on the basis of a damage image of a search target; and a search unit that searches for one or a plurality of pieces of first damage information similar to the second damage information from among the first damage information stored in the database on the basis of the second damage information acquired by the information acquisition unit.

Implementations described and claimed herein provide localization systems and methods using surface imaging. In one implementation, a raw image of a target surface is captured using at least one imager. The raw image is encoded into a template using at least one transform. The template specifies a course direction and an intensity gradient at one or more spatial frequencies of a pattern of the target surface. The template is compared to a subset of reference templates selected from a gallery stored in one or more storage media. A location of the target surface is identified when the template matches a reference template in the subset.