Abstract: Disclosed herein are an object recognition apparatus of an automated driving system using error removal based on object classification and a method using the same. The object recognition method is configured to train a multi-object classification model based on deep learning using training data including a data set corresponding to a noise class, into which a false-positive object is classified, among classes classified by the types of objects, to acquire a point cloud and image data respectively using a LiDAR sensor and a camera provided in an autonomous vehicle, to extract a crop image, corresponding to at least one object recognized based on the point cloud, from the image data and input the same to the multi-object classification model, and to remove a false-positive object classified into the noise class, among the at least one object, by the multi-object classification model.

Abstract: Aspects of the present disclosure are directed to, for example, a method for object detection, recognition, classification and tracking using a distributed networked architecture. In some embodiments, the distributed network architecture may include one or more sensor units wherein the image acquisition and the initial feature extraction are performed and a gateway processor for further data processing. Some aspects of the present disclosure are also directed to a vision system for object detection, and to algorithms implemented in the vision system for executing the method acts for object detection, recognition, classification and/or tracking.

Abstract: A method for clip selection includes receiving through an interface to a medical imaging device, a selection of a diagnostic procedure and a target portion of a mammalian body. Thereafter, the device acquires video clip imagery of the target portion and stores the video clip imagery in an image store. Each video clip of the video clip imagery is then image processed to determine a view and a quality of each video clip and a rule is retrieved from a rules base corresponding to the selected diagnostic procedure and target portion. In this regard, the rule specifies a requisite view and quality of the video clip imagery. Finally, the retrieved rule is applied to the video clip imagery as a filter to produce a subset of video clip imagery satisfying the specified requisite view and quality and the subset of video clip imagery is stored in the image store.

Abstract: A system uses facial recognition to exclude persons of interest (e.g., “undesirables”) from events and/or venues. Such a system can include a combination of cameras, edge processing devices, and servers that are in communications coupling (e.g., via a network) and that are on premise and/or in the cloud to recognize such persons of interest (POI) and interdict and prevent such POI from entering, traversing, and/or attending (collectively, “entering”) such venues or events.

Abstract: An image processing apparatus includes a memory configured to store first region detection information of a first frame; and a processor. The processor is configured to: identify a first pixel region corresponding to the first region detection information from a second frame, perform region growing processing on the first pixel region based on an adjacent pixel region that is adjacent to the first pixel region, obtain second region detection information of the second frame, based on the region growing processing, and perform image processing on the second frame based on the second region detection information.

Abstract: An image processing apparatus includes a processor configured to receive an image having a foreground segment and a background segment, receive a design object, calculate a position for superimposing the design object, on a basis of respective features of the foreground segment, the background segment, and the design object, and perform output to superimpose the design object at the calculated position.

Abstract: A method for providing diagnostic test results is provided. The method comprises providing a software application to be stored on a mobile device, the mobile device having a camera and a viewing screen, initiating operation of the camera, aligning the camera with a visual trigger associated with the diagnostic test, capturing an image of the diagnostic test, sending the image to a server, creating a pixel value array from the pixel values in the image, providing the pixel value array as inputs in a trained neural network, and providing either a positive or negative result from the trained neural network in response to the pixel value array.

Abstract: The present disclosure provides a multimodal fine-grained mixing method and system, a device, and a storage medium. The method includes: extracting data features from multimodal graphic and textual data, and obtaining each composition of the data features, the data features including a visual regional feature and a text word feature; performing fine-grained classification on modal information of each composition of the data features, to obtain classification results; and performing inter-modal and intra-modal information fusion on each composition according to the classification results, to obtain a fusion feature. The method enables a multimodal model to utilize a complementary characteristic of the multimodal data, with no influence by irrelevant information.

Abstract: An apparatus including a memory and a circuit. The memory may comprise three buffers. The circuit may be configured to allocate the three buffers in the memory based on a size of a full resolution feature map, receive a plurality of regions of interest ranked based on a feature map pyramid, generate a plurality of levels of the feature map pyramid starting from the full resolution feature map and store the levels in the buffers. The circuit may store the levels that are used by at least one of the plurality of regions of interest or do have a dependent level, the levels that are generated may be stored in the buffers in a pattern that ensures the level is stored until no longer needed to create the dependent level and enables the level to be discarded when no longer needed to create the dependent level.

Abstract: A computer-implemented method is provided for disentangled data generation. The method includes accessing, by a variational autoencoder, a plurality of supervision signals. The method further includes accessing, by the variational autoencoder, a plurality of auxiliary tasks that utilize the supervision signals as reward signals to learn a disentangled representation. The method also includes training the variational autoencoder to disentangle a sequential data input into a time-invariant factor and a time-varying factor using a self-supervised training approach which is based on outputs of the auxiliary tasks obtained by using the supervision signals to accomplish the plurality of auxiliary tasks.

Abstract: Disclosed are methods and systems for model-based data transformation.

Abstract: A video annotation system for deep learning based video analytics and corresponding methods of use and operation are described that significantly improve the efficiency of video data frame labeling and the user experience. The video annotation system described herein may be deployed at a network edge and may support various intelligent annotation functionality including annotation tracking, adaptive video segmentation, and execution of predictive annotation algorithms. In addition, the video annotation system described herein supports team collaboration functionality in connection with large-scale labeling tasks.

Abstract: The present disclosure provides a localization method as well as a helmet and a computer readable storage medium using the same. The method includes: extracting first feature points from a target image; obtaining inertial information of the carrier, and screening the first feature points based on the inertial information to obtain second feature points; triangulating the second feature points of the target image to generate corresponding initial three-dimensional map points, if the target image is a key frame image; performing a localization error loopback calibration on the initial three-dimensional map points according to at least a predetermined constraint condition to obtain target three-dimensional map points; and determining a positional point of the specific carrier according to the target three-dimensional map points. In this manner, the accuracy of the localization of a dynamic object such as a person when moving can be improved.

Abstract: Analysis for convolutional processing is performed using logic encoded in a semiconductor processor. The semiconductor chip evaluates pixels within an image of a person in a vehicle, where the analysis identifies a facial portion of the person. The facial portion of the person can include facial landmarks or regions. The semiconductor chip identifies one or more facial expressions based on the facial portion. The facial expressions can include a smile, frown, smirk, or grimace. The semiconductor chip classifies the one or more facial expressions for cognitive response content. The semiconductor chip evaluates the cognitive response content to produce cognitive state information for the person. The semiconductor chip enables manipulation of the vehicle based on communication of the cognitive state information to a component of the vehicle.

Abstract: An image processing method and apparatus, and a storage medium are provided. The method includes: obtaining a first image and a second image of a to-be-authenticated object, where the first image is captured by a first camera module, and the second image is captured by at least one second camera module; comparing the first image with image data in a target library for identity authentication, to obtain a first authentication result; and in response to that the first authentication result is authentication failure, performing joint authentication on the first image and the second image, and determining the identity of the to-be-authenticated object according to a second authentication result of the joint authentication.

Abstract: Systems and methods for classifying and/or sorting T cells by activation state are disclosed. The system includes a cell classifying pathway, a single-cell autofluorescence image sensor, a processor, and a non-transitory computer-readable memory. The memory is accessible to the processor and has stored thereon a trained convolutional neural network and instructions. The instructions, when executed by the processor, cause the processor to: a) receive the autofluorescence intensity image; b) optionally pre-process the autofluorescence intensity image to produce an adjusted autofluorescence intensity image; c) input the autofluorescence intensity image or the adjusted autofluorescence intensity image into the trained convolutional neural network to produce an activation prediction for the T cell.

Abstract: An image viewer and method for using the same in a medical image management system are described.

Abstract: In an embodiment, an apparatus includes: a sensor to sense real world information; a digitizer coupled to the sensor to digitize the real world information into digitized information; a signal processor coupled to the digitizer to process the digitized information into an image; a discriminator coupled to the signal processor to determine, based at least in part on the image, whether the real world information comprises an anomaly, where the discriminator is trained via a generative adversarial network; and a controller coupled to the discriminator.

Abstract: The invention relates to a device that receives images from one or more cameras, process images and automatically detects unknown humans in the field of view of the camera, for example to prevent burglary. In order to do so, the device comprises a processing logic configured to detect faces, recognize faces and verify if a face corresponds to a face in a collection of faces of known humans. If a face is detected, but does not correspond to a known face, an alarm event is triggered. The processing logic is further configured to classify objects in the image in classes of object comprising at least a human class. If a human is recognized, but no face has been detected for this human, an alarm event is also triggered. Thus, an alarm can be triggered in any case wherein a human is detected, which is not a known, trusty human.

Abstract: The application discloses a method for knowledge extraction based on TextCNN, comprising: S10, collecting first training data, and constructing a character vector dictionary and a word vector dictionary; S20, constructing a first convolutional neural network, and training the first convolutional neural network based on a first optimization algorithm, the first convolutional neural network comprises a first embedding layer, a first multilayer convolution, and a first softmax function connected in turn; S30, constructing a second convolutional neural network, and training the second convolutional neural network based on a second optimization algorithm, the second convolutional neural network comprises a second embedding layer, a second multilayer convolution, a pooling layer, two fully-connected layers and a second softmax function, the second embedding layer connected in turn; S40, extracting a knowledge graph triple of the to-be-predicted data according to an entity tagging prediction output by the first trai