Abstract: Disclosed herein is a system and method for improving the accuracy of an object detector when trained with a dataset having a significant number of missing annotations. The method uses a novel Background Recalibration Loss (BRL) which adjusts the gradient direction according to its own activation to reduce the adverse effect of error signals by replacing the negative branch of the focal loss with a mirror of the positive branch when the activation is below a confusion threshold.

Abstract: A method at a server for container location verification within a container yard, the method including requesting image data from at least one image sensor apparatus affixed to a container within the container yard; receiving the image data; and processing the image data to identify a location of a target container.

Abstract: There is provided a method of controlling a robot, which is performed by the robot that travels within a space. The robot identifies a person to be rescued, among persons within a space, by using a camera while travelling within the space, determines a distance between the robot and the person to be rescued based on depth information that is obtained by the camera and that is associated with the person to be rescued, and transmits information including the determined distance to a control server for the space.

Abstract: Devices, methods, and systems for occupancy detection are described herein. One device includes instructions to receive an image of a portion of a facility captured by an imaging device, the image defined by a field of view, and set a first occupancy detection threshold for a first part of the field of view and a second occupancy detection threshold for a second part of the field of view.

Abstract: A method for locating an optical code, on the basis of a finder pattern of the optical code, in an image having a plurality of pixels, the method comprising the steps of determining at least one candidate position for the finder pattern on the basis of edge transitions along at least one line through the image; determining a final position of the finder pattern in a vicinity of the candidate position; and verifying the final position, i.e. verifying that the image comprises the finder pattern in the final position, using a known property of the finder pattern.

Abstract: Systems and methods of authorizing access to access-controlled environments are provided. In one example, a method includes receiving, passively by a computing device, user behavior authentication information indicative of a behavior of a user of the computing device, comparing, by the computing device, the user behavior authentication information to a stored user identifier associated with the user, calculating, by the computing device, a user identity probability based on the comparison of the user behavior authentication information to the stored user identifier, receiving, by the computing device, a request from the user to execute an access-controlled function, and granting, by the computing device, the request from the user responsive to determining that the user identity probability satisfies a first identity probability threshold associated with the access-controlled function.

Abstract: A feature calculation means calculates N features for first data and N features for second data by using N feature functions for obtaining a feature for data on the basis of the data. A similarity degree calculation means calculates a similarity degree between the first data and the second data on the basis of the N features for the first data and the N features for the second data. Values of N features obtained when the same data is substituted into the N feature functions are different from each other.

Abstract: A method and system for fish identification based on body surface texture features and geometric features are provided. The method employs an improved Resnet network and a deep learning Yolov8 network to extract body surface texture features and geometric features of a fish on the basis of considering influences of fish tail swing and an oxygen concentration on a fish body form based on a small sample learning framework, and then realizes identity recognition of a fish individual by coupled analysis of the body surface texture features and the geometric features. The method can realize high-accuracy fish identification with few training samples of a fish to be identified from the perspective of actual application, provides theoretical basis and technical support for accurate fish stock assessment and accurate estimation of industrially farmed fish biomass, and meets the development requirements of modern agriculture.

Abstract: A smart table display system includes a display surface, an input/output device, and a processing circuit configured to receive financial data regarding a financial health of a user, generate a graphical user interface comprising a graphic that depicts the financial health of the user based on the financial data where the graphical user interface is for display on the display surface, receive sensor data regarding the user via the input/output device, continuously update the financial health of the user based on the sensor data, and cause the graphical user interface to update as the financial health of the user is updated.

Abstract: The invention relates to a method implemented by a computing device for processing math and text in handwriting, comprising: identifying symbols by performing handwriting recognition on a plurality of strokes; classifying, as a first classification, first symbols as either a text symbol candidate or a math symbol candidate with a confidence score reaching a first threshold; classifying, as a second classification, second symbols other than first symbols as either a text symbol candidate or a math symbol candidate with a respective confidence score by applying predefined spatial syntactic rules; updating or confirming, as a third classification, a result of the second classification by establishing semantic connections between symbols and comparing the semantic connections with the result of the second classification; and recognising each symbol as either text or math based on a result of said third classification.

Abstract: A scan driving circuit includes shift registers and clock signal lines. A shift register includes: an output circuit electrically connected to a scan input signal terminal and a pull-up node; a black frame insertion circuit electrically connected to a first clock signal terminal, a black frame insertion input signal terminal, a first voltage signal terminal, a second clock signal terminal and the pull-up node; and an output circuit electrically connected to the pull-up node, a third clock signal terminal, a shift signal terminal, a fourth clock signal terminal and a first output signal terminal. The shift registers include first shift registers and second shift registers. Third and fourth clock signal terminals of a first shift register are electrically connected to a same clock signal line. Third and fourth clock signal terminals of a second shift register are electrically connected to different clock signal lines.

Abstract: The present disclosure relates to systems and methods for processing real-time video and detecting objects in the video. In one implementation, a system is provided that includes an input port for receiving real-time video obtained from a medical image device, a first bus for transferring the received real-time video, and at least one processor configured to receive the real-time video from the first bus, perform object detection by applying a trained neural network on frames of the received real-time video, and overlay a border indicating a location of at least one detected object in the frames. The system also includes a second bus for receiving the video with the overlaid border, an output port for outputting the video with the overlaid border from the second bus to an external display, and a third bus for directly transmitting the received real-time video to the output port.

Abstract: A multi-layered moir target, useful in the calculation of misregistration between at least first, second and third layers being formed on a semiconductor device wafer, including at least one group of periodic structure stacks, each including a first stack, including a first stack first periodic structure (S1P1) having an S1P1 pitch along a first axis, a second stack, including a second stack first periodic structure (S2P1) having an S2P1 pitch along a second axis and a third stack, including a third stack first periodic structure (S3P1) having an S3P1 pitch along a third axis, the first axis being parallel to an x-axis or a y-axis, and at least one of the stacks including a second periodic structure having a second periodic structure pitch along at least one fourth axis parallel to the first axis and co-axial with one of the axes.

Abstract: The present disclosure relates to using end-to-end differentiable pipeline for optimizing parameters of a base procedural material to generate a procedural material corresponding to a target physical material. For example, the disclosed systems can receive a digital image of a target physical material. In response, the disclosed systems can retrieve a differentiable procedural material for use as a base procedural material in response. The disclosed systems can compare a digital image of the base procedural material with the digital image of the target physical material using a loss function, such as a style loss function that compares visual appearance. Based on the determined loss, the disclosed systems can modify the parameters of the base procedural material to determine procedural material parameters for the target physical material. The disclosed systems can generate a procedural material corresponding to the base procedural material using the determined procedural material parameters.

Abstract: Methods, apparatus, systems, and articles of manufacture are disclosed to decode receipts based on neural graph architecture. An apparatus includes interface circuitry to obtain an image of a document; machine readable instructions; and programmable circuitry to execute the machine readable instructions to at least generate nodes for a feature graph based on features extracted from text boxes, the nodes including polar coordinates indicative of angular positions of the text boxes; pass the feature graph through a graph neural network to generate an adjacency matrix; and identify text lines in the image of the document by clustering the nodes by line based on the adjacency matrix.

Abstract: An image processing system performs tilt correction with respect to a document image having handwritten characters and typed letters mixed with each other. The image processing system separates the document image into an image with handwritten characters determined as handwritten characters and an image without handwritten characters not determined as handwritten characters, estimates a tilt angle of the image without handwritten characters, and corrects the document image on the basis of the tilt angle.

Abstract: An apparatus comprising an interface and a processor. The interface may be configured to receive pixel data corresponding to a monitored area. The processor may be coupled to the interface and may be configured to (i) process the pixel data arranged as video frames, (ii) perform computer vision operations to detect an object in the video frames, (iii) extract characteristics about the object detected, (iv) generate a three-dimensional (3D) map of the object utilizing the pixel data corresponding to the object, (v) determine whether the object is within a three-dimensional (3D) virtual security zone based on the 3D map of the object, and (vi) generate a trigger signal in response to all or a portion of the object being within the 3D virtual security zone. The 3D virtual security zone is generally determined by a user during a setup procedure.

Abstract: The technologies described herein are generally directed to monitoring the configuration of an inventory of devices using captured images of a site. For example, a method described herein can include identifying a group of equipment installed at locations within a site. The method can further include, based on an image captured at the site by image capture equipment, detecting configuration activity at the site. Further, the method includes, based on analysis of the image, associating, by the tracking equipment, the configuration activity with equipment of the group of equipment.

Abstract: Provided herein is an integrated circuit including, in some embodiments, a special-purpose host processor, a neuromorphic co-processor, and a communications interface between the host processor and the co-processor configured to transmit information therebetween. The special-purpose host processor is operable as a stand-alone host processor. The neuromorphic co-processor includes an artificial neural network. The co-processor is configured to enhance special-purpose processing of the host processor through the artificial neural network. In such embodiments, the host processor is a keyword identifier processor configured to transmit one or more detected words to the co-processor over the communications interface. The co-processor is configured to transmit recognized words, or other sounds, to the host processor.

Abstract: A method for generating context information for a video stream includes selecting a set of frames of a video stream, applying a first machine learning model to the set of frames to extract action information from the set of frames, applying natural language learning to the set of frames to identify dialogue associated with the set of frames, and generating context information to categorize the dialogue and action information for the set of frames.

Abstract: From an input image comprising a portion of text, the portion of text is extracted. The portion of text comprises text in a first natural language. The input image is classified into a candidate scene within a set of predefined scenes. The candidate scene is scored according to a quality measure. Using a predefined scene parameter of the scored candidate scene, the portion of text is translated. In the input image, the portion of text is replaced. with the translated portion of text.

Abstract: A card-handling device may include a card intake, a card output, a card imaging device positioned between the card intake and the card output. The card imaging device may be configured to identify a non-conforming card. The card-handling device may be configured to store the non-conforming card in a designated location and/or to reorient the non-conforming card with a card-flipping apparatus configured to reorient flipped cards identified as non-conforming cards.

Abstract: Systems, methods, and computer programming products for estimating ingestion time of ingested files to be transformed into a searchable state for content mining by an on-premises computing environment or cloud environment, including multi-tenant cloud environments. Ingested files being indexed are analyzed for divisibility. Ingestion time varies based on the number of divisible elements (such as lines) of data within the ingested file and the amount of data per divisible element. A converter divides files into a plurality of elements treated as independent data and calculates the estimated ingestion time based on the number of divisions and file size for each divisible element. Estimated ingestion time is stored to internal fields corresponding to each divisible element in the index for the search data. During content mining, an internal condition is added to received search queries, displaying only search results where the estimated ingestion time is older than the current time.

Abstract: Cameras having storage fixtures within their fields of view are programmed to capture images and process clips of the images to generate sets of features representing product spaces and actors depicted within such images, and to classify the clips as depicting or not depicting a shopping event. Where consecutive clips are determined to depict a shopping event, features of such clips are combined into a sequence and transferred, along with classifications of the clips and a start time and end time of the shopping event, to a multi-camera system. A shopping hypothesis is generated based on such sequences of features received from cameras, along with information regarding items detected within the hands of such actors, to determine a summary of shopping activity by an actor, and to update a record of items associated with the actor accordingly.

Abstract: A method of configuring network elements in a design network topology includes receiving an image of the design network topology; attempting to retrieve design data from the received image corresponding to the design network topology; when the design data is retrieved, querying a topologies database using the design data to find a previously determined network topology that substantially matches the design network topology; when the design data is not retrieved, predicting a network topology using an unsupervised machine learning algorithm; identifying configurations for network elements in the matching network topology or in the predicted network topology in a configurations database; determining design configurations for the network elements of the design network topology from the identified configurations; translating the design configurations of the network elements to a standard format; and pushing the translated design configurations to actual network elements and/or virtual network elements correspondi

Abstract: A training data acquirer acquires training data including article image data, image-filter-related data indicating a combination of a plurality of image filters used for image processing of the article image data and a value of a parameter for each of the plurality of image filters, and optical character recognition (OCR) score data indicating a score of character recognition output through OCR when image processing is performed on the article image data using the image filters based on the image-filter-related data. A trained model generator generates a trained model indicating a relationship between the article image data, the image-filter-related data, and the OCR score data through machine learning using the training data.

Abstract: A method of x-ray projection geometry calibration in x-ray cone beam computed tomography, including: at least one step (S1) of obtaining two-dimensional x-ray images or a sinogram of at least a part of an object, generated through relatively rotating around the object a detector and an x-ray source projecting x-rays towards the detector; further including: at least one step (S4) of detecting in the two dimensional x-ray images or the sinogram at least one feature of the object by using a trained artificial intelligence algorithm; and at least one step of creating, based on the detection, calibration information which defines the geometry of the x-ray projection.

Abstract: The present invention provides a deep learning object detection method that locates the distant region in the image in real-time and concentrates on distant objects in a front dash cam perspective, trying to solve a common problem in advanced driver assistance system (ADAS) applications, that the detectable range of the system is not far enough.

Abstract: A computer device extracts an image feature of an image that includes one or more characters to be recognized. The image feature includes a plurality of image feature vectors. The device uses an attention mechanism to compute and output attention weight values corresponding to the target number of characters, based on the image feature vectors, through parallel computing. Each of the attention weight values corresponds to one or more respective characters and represents an importance of the plurality of image feature vectors for the respective characters. The device obtains at least one character according to the plurality of image feature vectors and the target number of attention weight values. Therefore, in a character recognition process, with recognition based on the foregoing attention mechanism, a character in any shape can be effectively recognized by using a simple procedure, thereby avoiding a cyclic operation process and greatly improving operation efficiency.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for computer aided diagnosis of a medical image. One of the methods includes processing the medical image through a machine learning (ML) model to provide a first feature representation of the medical image, wherein the ML model includes an input layer and a pooling layer, wherein the first feature representation is an output from the pooling layer, generating, by the ML model, a sequence of second feature representations of the medical image from the first feature representation of the medical image, wherein each second feature representation in the sequence of the second feature representations having a lower dimension than the first feature representation, and generating, by the ML model, an output as a last second feature representation in the sequence of the second feature representations.

Abstract: In accordance with implementations of the present disclosure, there is provided a solution for portrait editing and synthesis. In this solution, a first image about a head of a user is obtained. A three-dimensional head model representing the head of the user is generated based on the first image. In response to receiving a command of changing a head feature of the user, the three-dimensional head model is transformed to reflect the changed head feature. A second image about the head of the user is generated based on the transformed three-dimensional head model, and reflects the changed head feature of the user. In this way, the solution can realize editing of features like a head pose and/or a facial expression based on a single portrait image without manual intervention and automatically synthesize a corresponding image.

Abstract: To easily and correctly determine substantial identity of a plurality of images in each of which an object is placed, provided is an image judgement method including steps of obtaining first object data from a first image with a use of an R-CNN 12, which is a first machine learning model, where the first object data indicates an attribute and a layout of an object in the first image, obtaining second object data from a second image with a use of the R-CNN 12, where the second object data indicates an attribute and a layout of an object in the second image, and determining substantial identity of the first image and the second image based on the first object data and the second object data with a use of a CNN, which is a second machine learning model.

Abstract: A method of processing input data for a given layer of a neural network using a data processing system comprising compute resources for performing convolutional computations is described. The input data comprises a given set of input feature maps, IFMs, and a given set of filters. The method comprises generating a set of part-IFMs including pluralities of part-IFMs which correspond to respective IFMs, of the given set of IFMs. The method further includes grouping part-IFMs in the set of part-IFMs into a set of selections of part-IFMs. The method further includes convolving, by respective compute resources of the data processing system, the set of selections with the given set of filters to compute a set of part-output feature maps. A data processing system for processing input data for a given layer of a neural network is also described.

Abstract: An object recognition method, an object recognition system, and a readable storage medium are provided. The object recognition method is to recognize an object from a first image, and includes: acquiring a first image; calculating depth information of the first image; performing superpixel segmentation on the first image, to obtain a superpixel image; generating three-dimensional image data of the first image according to the depth information and the image data of the superpixel image; and inputting the three-dimensional image data into a depth neural network for object recognition, to obtain a recognition result.

Abstract: A method of tracking an object includes obtaining a likelihood of a free behavior model of the object and a likelihood of a constant speed model of the object using the position, the speed, and the type of the object determined at a previous time point and the position, the speed, and the type of the object determined at the current time point, and correcting the type of the object at the current time point using the likelihood of the free behavior model, the likelihood of the constant speed model, and the measured type of the object.

Abstract: An image processing apparatus includes one or more memories storing instructions, and one or more processors executing the instructions to obtain document data before printing, to set an inspection area that is an inspection target for an image represented by the obtained document data, to update, in a case when the document data is modified, the set inspection area based on the modified document data, and to generate, based on the modified document data, a reference image for inspection of a printed material.

Abstract: An image recognition method and apparatus. The method comprises: obtaining original image data, convolutional neural network configuration parameters, and convolutional neural network operation parameters from a data transfer bus, the original image data comprising M pieces of pixel data, and M being a positive integer (101); and performing convolutional neural network operation on the original image data by a convolutional neural network operation module according to the convolutional neural network configuration parameters and the convolutional neural network operation parameters (102), wherein the convolutional neural network operation module comprises a convolution operation unit, a batch processing operation unit, and an activation operation unit connected in sequence. The method improves the real timeliness of image recognition.

Abstract: A firewall system stores filtering criteria which include rules for blocking presentation of all or a portion of the media content based at least in part on an identity of an individual appearing in the media content. The firewall system receives the media content. The firewall system determines the identity of the individual appearing in the media content. Based at least in part on the identity of the individual appearing in the media content and the filtering criteria, the firewall system determines an action for securing the media content. The action may be allowing presentation of the media content, blocking presentation of the media content, or blocking presentation of a portion of the media content. The determined action is automatically implemented.

Abstract: The disclosure relates to a method for flagging at least an event of interest in an unlabeled time series of a parameter relative to a wellsite (including to the well, formation or a wellsite equipment), wherein the time series of the parameter is a signal of the parameter as a function of time. The disclosure also relates to a method for evaluation a downhole operation such as a pressure test using a pressure time series. Such methods comprises collecting a time series, extracting at least an unlabeled subsequence of predetermined duration in the time series, and assigning an event of interest a label, in particular representative of the status of the downhole operation, to at least one of the unlabeled subsequences. A command may be sent to a wellsite operating system based on assigned label.

Abstract: Methods and systems for automated construction of an anonymized facial recognition library are disclosed. A camera of a client device may capture a first plurality of images of faces of members of a panel of viewers of media content presented on a content presentation device collocated with the client device during viewing sessions. A first machine learning (ML) model may be applied to the first plurality to generate a second plurality of feature vectors, each associated with a different one of the images. One or more clusters of feature vectors of the second plurality may be computationally determined within a vector space of the feature vectors. A respective centroid feature vector may be determined for each respective cluster, and assigned a unique ID. A respective association between each cluster ID and a respective name ID may be determined based on panel-member information received at the client device.

Abstract: Described is a method that involves operating an unmanned aerial vehicle (UAV) to begin a flight, where the UAV relies on a navigation system to navigate to a destination. During the flight, the method involves operating a camera to capture images of the UAV's environment, and analyzing the images to detect features in the environment. The method also involves establishing a correlation between features detected in different images, and using location information from the navigation system to localize a feature detected in different images. Further, the method involves generating a flight log that includes the localized feature. Also, the method involves detecting a failure involving the navigation system, and responsively operating the camera to capture a post-failure image. The method also involves identifying one or more features in the post-failure image, and determining a location of the UAV based on a relationship between an identified feature and a localized feature.

Abstract: The present disclosure generally relates to methods and user interfaces for managing visual content at a computer system. In some embodiments, methods and user interfaces for managing visual content in media are described. In some embodiments, methods and user interfaces for managing visual indicators for visual content in media are described. In some embodiments, methods and user interfaces for inserting visual content in media are described. In some embodiments, methods and user interfaces for identifying visual content in media are described. In some embodiments, methods and user interfaces for translating visual content in media are described. In some embodiments, methods and user interfaces for translating visual content in media are described. In some embodiments, methods and user interfaces for managing user interface objects for visual content in media are described.

Abstract: Embodiments of the present disclosure relate to a method, an electronic device, and a computer program product for training a data classification model. The method includes generating a first training rule based on probabilities of classifying a plurality of sample data into corresponding classes by a data classification model. The method also includes generating a second training rule based on relevances of the plurality of sample data to the corresponding classes. In addition, the method also includes training the data classification model using the first training rule and the second training rule. With this method, a data classification model is trained, so that the data classification accuracy of the data classification model and the robustness to noise can be improved.

Abstract: A messaging system performs neural network hair rendering for images provided by users of the messaging system. A method of neural network hair rendering includes processing a three-dimensional (3D) model of fake hair and a first real hair image depicting a first person to generate a fake hair structure, and encoding, using a fake hair encoder neural subnetwork, the fake hair structure to generate a coded fake hair structure. The method further includes processing, using a cross-domain structure embedding neural subnetwork, the coded fake hair structure to generate a fake and real hair structure, and encoding, using an appearance encoder neural subnetwork, a second real hair image depicting a second person having a second head to generate an appearance map. The method further includes processing, using a real appearance renderer neural subnetwork, the appearance map and the fake and real hair structure to generate a synthesized real image.

Abstract: A method and related system operations include obtaining a video stream with an image sensor of a camera device, detecting a plurality of target objects by executing a neural network model based on the video stream with a vision processor unit of the camera device. The method also includes generating a plurality of bounding boxes, determining a plurality of character sequences by, for each respective bounding box of the plurality of bounding boxes, performing a set of optical character recognition (OCR) operations to determine a respective character sequence of the plurality of character sequences. The method also includes updating a plurality of tracklets to indicate the plurality of bounding boxes and storing the plurality of tracklets in association with the plurality of character sequences in a memory of the camera device.

Abstract: A monitoring system is configured to monitor a property. The monitoring system includes a camera, a sensor, and a monitor control unit. The monitor control unit is configured to receive image data and sensor data. The monitor control unit is configured to determine that the image data includes a representation of a person. The monitor control unit is configured to determine an orientation of a representation of a head of the person. The monitor control unit is configured to determine that the representation of the head of the person likely includes a representation of a face of the person. The monitor control unit is configured to determine that the face of the person is likely concealed. The monitor control unit is configured to determine a malicious intent score that reflects a likelihood that the person has a malicious intent. The monitor control unit is configured to perform an action.

Abstract: A method can include receiving (1) images of at least one subject and (2) at least one total mass value for the at least one subject. The method can further include executing a first machine learning model to identify joints of the at least one subject. The method can further include executing a second machine learning model to determine limbs of the at least one subject based on the joints and the images. The method can further include generating three-dimensional (3D) representations of a skeleton based on the joints and the limbs. The method can further include determining a torque value for each limb, based on at least one of a mass value and a linear acceleration value, or a torque inertia and an angular acceleration value. The method can further include generating a risk assessment report based on at least one torque value being above a predetermined threshold.

Abstract: At least a main area and a candidate area are provided on a display screen of a display device, and a display control unit that causes content displayed in the candidate area when reproduction of content displayed in the main area ends to be displayed in the main area, and a detection unit that detects a person from a captured image of at least a place where the display screen is visually recognizable are included, wherein the display control unit changes the content displayed in the candidate area depending on the person during reproduction of the content displayed in the main area.

Abstract: In one implementation, a method of defining a negative space in a three-dimensional scene model is performed at a device including a processor and non-transitory memory. The method includes obtaining a three-dimensional scene model of a physical environment including a plurality of points, wherein each of the plurality of points is associated with a set of coordinates in a three-dimensional space. The method includes defining a subspace in the three-dimensional space with less than a threshold number of the plurality of points. The method includes determining a semantic label for the subspace. The method includes generating a characterization vector of the subspace, wherein the characterization vector includes the spatial extent of the subspace and the semantic label.

Abstract: A feature mapping computer system configured to (i) receive a localized image including a photo depicting a driving environment and location data associated with the photo, (ii) identify, using an image recognition module, a roadway feature depicted in the photo, (iii) generate, using a photogrammetry module, a point cloud based upon the photo and the location data, wherein the point cloud comprises a set of data points representing the driving environment in a three dimensional (“3D”) space, (iv) localize the point cloud by assigning a location to the point cloud based upon the location data, and (v) generate an enhanced base map that includes a roadway feature.