Abstract: An information processing apparatus includes a processor configured to: extract a first region showing a feature of an object to be registered and having a predetermined size from an object surface image obtained by imaging a surface of the object; specify, from the object surface image, a second region to be compared with the first region and having a same size as the first region; derive a similarity between the first region and the second region; and register, as a registered image, an image of the first region in a case where the similarity between the first region and the second region satisfies a predetermined standard.

Abstract: An information processing device (100) includes an image acquisition unit (110), a measurement unit (172), and a recognition processing unit (173). The image acquisition unit (110) acquires an image by receiving external light by a light receiving unit. The measurement unit (172) measures a distance to a subject by using a first image based on the image acquired by the image acquisition unit in response to emission of infrared light from a light emitting unit. The recognition processing unit (173) performs subject recognition processing by using a second image based on an image acquired by the image acquisition unit (110) that has received external light without infrared light emitted from the light emitting unit.

Abstract: A system and method for pharmacophore-conditioned generation of molecules. The system and method modifies a conditional variational autoencoder (CVAE) such that the latent space in generation of a molecule is not conditioned on the pharmacophore space of the molecule. This allows for generation of pharmacophore descriptors independently from the conditional on which CVAE has been trained, removing a substantial impediment to the use of CVAEs for exploration of pharmacophore descriptors of a molecule.

Abstract: A method, an apparatus and a non-transitory computer readable medium may be provided. In one aspect, a method performed by a computer includes: matching an input image against at least one of images corresponding to a network of subjects to identify a subject corresponding to the input image, the input image comprising at least one label corresponding to the identified subject; and determining whether or not to eliminate a node representing the subject in the network of subjects based on the at least one corresponding label.

Abstract: A non-transitory computer-readable recording medium has stored therein an information processing program causes a computer to execute a process comprising, dividing input data including a plurality of items related to data to be estimated that is a target of an estimation process using a machine learning model into a plurality of groups based on a predetermined condition; acquiring, for each of the plurality of groups, a first estimation result by a machine learning model that has already been trained, based on the input data included in the each of the plurality of groups; and outputting estimation result information indicating a basis for estimation of a second estimation result that is an estimation result of the data to be estimated, based on the first estimation result of each of the plurality of groups and the second estimation result.

Abstract: Described herein are systems and methods for generating multi-class machine learning classifiers using a plurality of hyper-trained binary classifiers to perform both characteristic detection and object type classification. A range associated with a characteristic of the data sought to be classified may be divided into a plurality of sub-ranges that can be ordered. A plurality of binary machine learning classifiers can be generated, with each binary classifier associated with one of the sub-ranges. Each binary classifier of the plurality of classifiers may be generated by training the classifier to determine if a given input data set contains a characteristic within or outside of the sub-range associated with the binary classifier. Once the binary classifiers have been trained, they can be used to determine what value within the range of the characteristic an input data sample has, and may be used to determine other properties about the input data.

Abstract: The present disclosure belongs to the technical field of artificial intelligence, and discloses an automatic classification method and system of teaching videos based on different presentation forms, which with three convolutional neural network models, may accurately locate the information required for teaching video classification by two self-trained YOLOV4 target detection neural network models and human body key point detection technology, solves the problem that the background and character features of teaching videos do not change significantly, and improves the accuracy of feature extraction. The structure of the self-built convolutional neural network models is suitable for classification of Interview type and Head type teaching videos. The depth of the network is just appropriate compared with several classical video classification algorithms, which reduces the energy consumption of computer hardware.

Abstract: An information processing apparatus includes at least one memory storing a program, and at least one processor which, by executing the program, causes the information processing apparatus to acquire learning image data, ground truth region image data which indicates a ground truth region of the first region included in the learning image data, and ground truth data on the recognition, acquire estimated region image data which indicates an estimated region of the first region in the learning image data, and train the learning model using a first training dataset constituted of the learning image data, the ground truth region image data and the ground truth data, and a second training dataset constituted of the learning image data, the estimated region image data and the ground truth data.

Abstract: Disclosed are an image classification method performed by a computer device. The method includes: acquiring an image feature of a pathological image; extracting, for each scale in multiple scales, a local feature corresponding to the scale from the image feature; splicing the local features respectively corresponding to the scales to obtain a spliced image feature; and classifying the spliced image feature to obtain a category to which the pathological image belongs. According to the method provided in the embodiments of this application, the local features corresponding to different scales contain different information, so that the finally obtained spliced image feature contains feature information corresponding to different scales, and the feature information of the spliced image feature is enriched. The category to which the pathological image belongs is determined based on the spliced image feature, so that the accuracy of the category is ensured.

Abstract: An electronic device is provided. The electronic device includes a communication circuit configured to transmit and receive data to and from an external device, a memory, and a processor, wherein the processor may store a user pattern in the memory based on first sensing information detected during a first time period by at least one of an external camera device or an external Internet of things (IoT) device, detect occurrence of an irregular event based on second sensing information detected during a second time period by the external camera device or the external IoT device and the stored user pattern, match the irregular event that occurs with a plurality of images photographed by the external camera device, and determine priorities of the plurality of images based on a matching result.

Abstract: An unmanned aerial vehicle based system, method, and computer program product, for inspection and monitoring construction progress of outdoor power facilities. The method can receive a construction plan and schedule identifying milestones for installation at an outdoor construction site of at least one externally placed piece of equipment specified in the construction plan. A machine learning artificial intelligence processing system identifies, from a set of observed aerial images of the outdoor construction site captured by an unmanned aerial vehicle, at least one externally placed piece of equipment in the outdoor construction site. The identified piece of equipment in the construction site is compared to an expected piece of equipment specified in the construction plan. The difference between identified and expected piece of equipment is reported and the construction schedule can be updated based on the difference. An alert is provided if the difference exceeds a threshold.

Abstract: Computer-implemented methods and systems are provided for pre-processing image data to generate improved training data for training a machine learning system to automatically classify imagery. One or more images of a geographic region are captured and processed to obtain a plurality of labelled samples (also referred to as patches) for training the machine learning system. Training data can then be selected from these labelled samples using a method based on simulated annealing, which iteratively searches through subsets of the available patches to identify a solution that is as close to the most optimal solution as possible.

Abstract: A non-transitory computer-readable recording medium stores an image processing program for causing a computer to execute a process including: obtaining a plurality of images consecutively captured in a time series manner; calculating a probability that a type of an object present in each of the plurality of images is one type using a trained classification model; determining whether or not the probability that the type of the object is the one type periodically changes in consecutive images among the plurality of images; and in a case of determining that the probability periodically changes, saving one image within a period in which the probability that the type of the object is the one type periodically changes as training data.

Abstract: An image processing apparatus has a first image acquisitor that acquires a source image, a second image acquisitor that acquires a first target image, a label acquisitor that acquires a label, a feature extractor including a first neural network that extracts a feature of the source image and a feature of the first target image, a class classifier including a second neural network that performs a class classification of the source image and the first target image, a domain classifier including a third neural network that performs a domain classification of the source image and the first target image, a processor that assigns a pseudo label to the first target image, a self-learner that performs a self-learning of the first neural network, the second neural network, and the third neural network, and a learner that learns the first, second and third neural networks, by performing a back propagation process.

Abstract: Provided are a method for training a region of interest (ROI) detection model, a method for detecting an ROI, a device, and a medium. The specific implementation includes: performing feature extraction on a sample image to obtain a sample feature data; performing non-linear mapping on the sample feature data to obtain a first feature data and a second feature data; determining an inter-region difference data according to the second feature data and a third feature data of the first feature data in a region associated with a label ROI; and adjusting at least one of a to-be-trained feature extraction parameter and a to-be-trained feature enhancement parameter of the ROI detection model according to the inter-region difference data and the region associated with the label ROI.

Abstract: There are provided a negative example availability deciding apparatus, a negative example availability deciding method, and a program that enable reuse of training data. A positive example evaluation training section generates a positive example evaluation classifier trained with target data as a positive example. A negative example evaluation training section generates a negative example evaluation classifier trained with target data as a negative example. An availability deciding section determines the classification accuracy of the positive example evaluation classifier and the classification accuracy of the negative example evaluation classifier by using evaluation data. The availability deciding section decides, on the basis of the classification accuracy of the positive example evaluation classifier and the classification accuracy of the negative example evaluation classifier, whether or not to use target data as negative example training data for training another classifier.

Abstract: An de-identification method includes: performing sampling on a target space of a projector in a target network model to obtain N virtual identity vectors, N being a positive integer; extracting, by using an attribute extractor in the target network model, M attribute vectors from a target image, M being a positive integer; and generating, by using a generator in the target network model, a de-identified image of the target image based on the N virtual identity vectors and the M attribute vectors.

Abstract: A learning apparatus includes: an extraction unit that extracts a first false recognition target which is a recognition target in regard to which a probability that a first recognition device falsely recognizes it as a first recognition target from among a plurality of types of recognition targets is equal to or greater than a predetermined percentage; a training data generation unit that generates first training data that includes an image including both the first recognition target and the first false recognition target; and a learning control unit that makes, by using the first training data, a first recognition device learn that the first false recognition target is a recognition target of a type different from that of the first recognition target.

Abstract: Disclosed are a table recognition method and apparatus. The table recognition method includes steps of obtaining an image vision feature and a character content feature of a table image; fusing the image vision feature and the character content feature of the table image to acquire a first fusion feature, and carrying out recognition based on the first fusion feature to acquire a table structure; and performing, based on the table structure, character recognition on the table image to acquire table character contents.

Abstract: Example methods, apparatuses, and/or articles of manufacture are disclosed that may be implemented, in whole or in part, techniques to process pixel values sampled from a multi color channel imaging device. In particular, methods and/or techniques to process pixel samples for interpolating pixel values for one or more color channels.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for training an object recognition neural network using multiple data sources. One of the methods includes receiving training data that includes a plurality of training images from a first source and images from a second source. A set of training images are obtained from the training data. For each training image in the set of training images, contrast equalization is applied to the training image to generate a modified image. The modified image is processed using the neural network to generate an object recognition output for the modified image. A loss is determined based on errors between, for each training image in the set, the object recognition output for the modified image generated from the training image and ground-truth annotation for the training image. Parameters of the neural network are updated based on the determined loss.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for selecting actions to be performed by an agent interacting with an environment to accomplish a goal. In one aspect, a method comprises: obtaining an observation characterizing a state of the environment, processing the observation using an embedding model to generate a lower-dimensional embedding of the observation, determining an auxiliary task reward based on a value of a particular dimension of the embedding, determining an overall reward based at least in part on the auxiliary task reward, and determining an update to values of multiple parameters of an action selection neural network based on the overall reward using a reinforcement learning technique.

Abstract: A method and system of multiple facial attributes recognition using highly efficient neural networks.

Abstract: Apparatuses, systems, and techniques to generate a robust representation of an image. In at least one embodiment, input tokens of an input image are received, and an inference about the input image is generated based on a vision transformer (ViT) system comprising at least one self-attention module to perform token mixing and a channel self-attention module to perform channel processing.

Abstract: Systems, methods, and devices for detecting brake lights are disclosed herein. A system includes a mode component, a vehicle region component, and a classification component. The mode component is configured to select a night mode or day mode based on a pixel brightness in an image frame. The vehicle region component is configured to detect a region corresponding to a vehicle based on data from a range sensor when in a night mode or based on camera image data when in the day mode. The classification component is configured to classify a brake light of the vehicle as on or off based on image data in the region corresponding to the vehicle.

Abstract: The present technology relates to an information processing apparatus, an information processing method, and a program that enable processing with maintained accuracy to be performed using a recognition model according to a noise level. A model switching unit that switches a recognition model according to an amount of noise of an imaged image is provided, and the model switching unit switches between a first recognition model trained with an image without noise and a second recognition model trained with an image with noise. The recognition model is used for processing by an AI function. It is applicable to, for example, a system including a surveillance camera and a server that distributes data from the surveillance camera.

Abstract: Presented herein are systems and methods for identifying and disambiguating individual objects of interest from a data set. In one or more examples, image data can be received and processed to identify candidate images that may contain one or more objects of interest. The candidate image data can then be processed to segment potential objects of interest from the candidate images. The segmented potential objects of interest can then be processed via one or more analytics to determine whether each potential objects of interest is an object of interest, to determine an object type, and/or to disambiguate specific objects of interest.

Abstract: An image based valuation system to perform operations that include: causing display of image data at a client device, the image data comprising a depiction of an object that comprises a set of object features; detecting the object based on the depiction of the object within the image data, the object corresponding with an object class; identifying the set of object features that correspond with the object; accessing a repository that corresponds with the object class, the repository comprising at least an indication of a valuation of the object based on the set of object features; and causing display of a presentation of the valuation at the client device.

Abstract: Systems and methods are described for disambiguating user input based on a physical location of items in a vicinity of a user. The system determines that a query received from a user contains an ambiguity. In response, the system identifies several items in the physical vicinity of the user. Then, the system analyzes the identified plurality of items to determine whether the plurality of items forms a first ensemble of items or a second ensemble of items. If the plurality of items forms a first ensemble of items, the system performs a search using the search query and a first keyword related to the first ensemble of items. If the plurality of items forms a second ensemble of items, the system performs a search using the search query and a second keyword related to the second ensemble of items. The system then outputs results of the performed search.

Abstract: A catastrophe analysis computing system includes a processor; an electronic network; and a memory having stored thereon non-transitory computer-executable instructions that, when executed by the processor, cause the system to: cause a client device to receive a exposure map layer; cause a filter graphical user interface and wind speed layers to be displayed in a graphical user interface; receive an input; and update the map graphical user interface. In another aspect, a method includes causing a client device to receive an exposure map layer; causing a filter graphical user interface and wind speed layers to be displayed in a graphical user interface; receiving an input; and updating the map graphical user interface. A method includes receiving an exposure map; displaying a filter graphical user interface; receiving an input; and updating a map graphical user interface.

Abstract: An embodiment behavior data augmenting apparatus includes a memory storing algorithms and data and a processor configured to execute the algorithms stored in the memory to extract an object region from video data, define a spatiotemporal characteristic for each class of behavior data by a behavior of an object in the object region, augment the behavior data, and perform learning to recognize the behavior of the object based on the augmented behavior data and a learning algorithm.

Abstract: Systems and methods for assessing performance of a player or a team comprising a plurality of players are described. An example method includes receiving a pass distribution matrix corresponding to a number of completed passes within a field of play for one or more players of the plurality of players, selecting a particular player from the one or more players, and generating a spine for the team. In an example, generating the spine for the team includes producing a visual representation of the field of play that illustrates a position of the particular player and at least one other player within the field of play, and automatically varying, on the visual representation, a thickness of each of a plurality of connecting lines between the particular player and the at least one other player based on the number of completed passes for the particular player.

Abstract: A system and a method of an automatic video production based on an off-the-shelf video camera are provided herein. The method may include receiving, from an off-the-shelf camera, a footage of video image frames containing a scene of a sport event, uploading the footage to a computing device, and processing, by the computing device, the footage to automatically generate a video production of the sport event based on at least a portion of the video frame images.

Abstract: A system may receive video information. The system may select a combination of light-weight features and heavy weight features. The light-weight features may be extracted from the video information and the heavy weight features not extracted. The system may forecast, based on the light-weight features and the heavy weight features, accuracy, and latency metrics for performing the object detection and tracking using a plurality of candidate branch configurations, respectively. The system may select a branch configuration from the plurality of candidate branch configurations in response to satisfaction of an optimization criterion. The system may perform object detection and tracking based on the selected branch configuration.

Abstract: Techniques are described that support automated generation of a digital document from digital videos using machine learning. The digital document includes textual components that describe a sequence of entity and action descriptions from the digital video. These techniques are usable to generate a single digital document based on a plurality of digital videos as well as incorporate user-specified constraints in the generation of the digital document.

Abstract: Methods and systems for automated video segmentation are disclosed. A sequence of video frames having video segments of contextually-related sub-sequences may be received. Each frame may be labeled according to segment and segment class. A video graph may be constructed in which each node corresponds to a different frame, and each edge connects a different pair of nodes, and is associated with a time between video frames and a similarity metric of the connected frames. An artificial neural network (ANN) may be trained to predict both labels for the nodes and clusters of the nodes corresponding to predicted membership among the segments, using the video graph as input to the ANN, and ground-truth clusters of ground-truth labeled nodes. The ANN may be further trained to predict segment classes of the predicted clusters, using the segment classes as ground truths. The trained ANN may be configured for application runtime video sequences.

Abstract: The present invention relates to an independent toilet seat assembly with image recognizers for automatic lifts, which Includes a seat assembly having a cushion ring and one upper lid extended therefrom and at least one image recognizer. The seat assembly can be combined on a toilet pot, and it has provided with one controller and two-powered units inside. The image recognizer can be located on the side of the seat assembly, which is electrical connected to the controller, or installed on the wall at the back of the toilet at an appropriate position, which is electrically connected to the controller by wireless means, such as a bluetooth, etc. The controller is connected to two power units, wherein the first power unit is used to control the automatic lifting and closing of the cover lid. The second power unit is used to control the automatic lifting and closing of the cushion ring.

Abstract: Methods and apparatuses using a cooking history are provided. An electronic apparatus includes a memory storing instructions, a plurality of cooking histories, and a plurality of cooking objects, and a processor configured to execute the instructions to, based on identification information regarding the plurality of cooking objects, identify a cooking object corresponding to the identification information from among the plurality of cooking objects; based on cooking setting information corresponding to a user input, identify a cooking history corresponding to the cooking setting information from among the plurality of cooking histories; obtain a cooking prediction result corresponding to the cooking history; and provide, to a user of the electronic apparatus, information regarding the cooking prediction result.

Abstract: Monitoring systems and methods for identifying an object of interest after the object of interest has undergone a change in appearance. One example provides an image sensor is configured to monitor a first area. A first electronic processor is configured to detect a first appearance of an object of interest within the first area, and determine a visual characteristic of the object of interest. The first electronic processor is configured to receive a first notification indicative of movement of the object of interest into a second area and an access input, and associate the visual characteristic of the object of interest with the access input. The first electronic processor is configured to detect a second appearance of the object of interest within the first area, and update the visual characteristic of the object of interest based on the second appearance of the object of interest.

Abstract: To efficiency search for an object associated with a sensed event, an information processing apparatus includes a sensor that analyzes a captured video and senses whether a predetermined event has occurred, a determining unit that determines a type of an object to be used as query information based on a type of the event in response to sensing of the event occurrence, and a generator that detects the object of the determined type from the video and generates the query information based on the detected object.

Abstract: ADAS includes a processing circuit and a memory which stores instructions executable by the processing circuit. The processing circuit executes the instructions to cause the ADAS to receive, from a vehicle that is in motion, a video sequence, generate a position image including at least one object included in the stereo image, generate a second position information associated with the at least one object based on reflected signals received from the vehicle, determine regions each including at least a portion of the at least one object as candidate bounding boxes based on the stereo image and the position image, and selectively adjusting class scores of respective ones of the candidate bounding boxes associated with the at least one object based on whether a respective first position information of the respective ones of the candidate bounding boxes matches the second position information.

Abstract: Disclosed herein are system, method, and computer program product aspects for enabling an autonomous vehicle (AV) to detect objects and forecast their predicted positions. The system can monitor an object within a vicinity of the AV. A plurality of trajectories predicting paths the object will take at a future time can be generated, the plurality of trajectories being based on a generated three-dimensional (3D) point cloud map indicating current and past characteristics of the object. Using a learned model, a forecasted position of the object at an instance in time can be generated along one or more of the plurality of trajectories. A maneuver for the AV can be performed based on the forecasted position.

Abstract: A control device includes circuitry configured to: obtain an external environment recognition image representing a recognition result of a surrounding area of a moving body; obtain obstacle information representing a detection result of an obstacle in the surrounding area of the moving body; and display the obstacle information on a display device in a superimposed manner on the external environment recognition image. The obstacle information indicates a region in which the obstacle is detected among a plurality of fan-shaped detection regions centered at the moving body when viewed from above the moving body, and a size of a central angle of the fan-shaped detection region is set according to a distance from the moving body.

Abstract: A method for determining a safe state for a vehicle includes disposing a camera at a vehicle and disposing an electronic control unit (ECU) at the vehicle. Frames of image data are captured by the camera and provided to the ECU. An image processor of the ECU processes frames of image data captured by the camera. A condition is determined via processing at the image processor of the ECU frames of image data captured by the camera. The condition includes a shadow present in the field of view of the camera within ten frames of image data captured by the camera or a damaged condition of the imager within two minutes of operation of the camera. The ECU determines a safe state for the vehicle responsive to determining the condition.

Abstract: A method for displaying lane information on an augmented reality display includes receiving roadway data. The roadway data includes information about a roadway along a route of a vehicle. The roadway includes a plurality of lanes. The roadway data includes lane information about at least one of the plurality of lanes along the route of the vehicle. The method further includes receiving vehicle-location data. The vehicle-location data indicates a location of the vehicle. The method further includes determining that that the vehicle is approaching a road junction using the vehicle-location data and the roadway data. The method further includes, in response to determining that the vehicle is approaching the road junction, transmitting a command signal to a dual-focal plane augmented reality display to display at least one virtual image that is indicative of the lane information.

Abstract: Provided are a computer-implemented method and apparatus for predicting virtual road sign locations of virtual road signs that may be superimposed onto environmental data of a vehicle. The method includes collecting, as a first training data subset, one or more aerial and/or satellite images of a pre-determined region; obtaining, as a second training data subset, geocentric positions of key point markers in the pre-determined region; supplying the first and second training data subsets to a deep neural network as training dataset; training the deep neural network on the training dataset to predict key point marker locations in a region of interest, the key point marker locations corresponding to virtual road sign locations; defining a region of interest as input dataset; and processing the input dataset by the trained deep neural network to predict key point marker locations within the defined region of interest.

Abstract: Disclosed herein is a driver assistance apparatus including a camera installed on a vehicle, having a forward field of view of the vehicle, and configured to acquire image data and a processor configured to process the image data. The processor may acquire image data of road guide signs from the image data and identify a position of the vehicle based on the image data of the road guide signs.

Abstract: There is provided methods and apparatuses for detecting a lane using a light detection and ranging (LiDAR), the apparatus including a processor configured to generate a range image based on a LiDAR point cloud acquired from the LiDAR, and acquire at least one lane data by detecting at least one lane area present in the range image using a trained lane detection model.

Abstract: A stop line detection apparatus recognizes a boundary line of a lane on which the host vehicle travels, and detect a stop line at an intersection in front of the host vehicle. The apparatus sets an intersection area corresponding to the intersection in the road on which the host vehicle travels on the basis of a captured image by a front camera of the host vehicle, detects a crosswalk in the intersection area, a stop line search area setting unit configured to set a stop line search area extending from the crosswalk toward a front side in the lane on which the host vehicle travels, and performs stop line detection in the stop line search area. A depth of the stop line search area in an extending direction of a traveling lane is smaller than twice a depth of the crosswalk.

Abstract: According to an embodiment, a reading system includes a reader and a calculator. The reader reads, from a character image, a character that is displayed by a segment display. The calculator performs one of first, second, third, or fourth processing. In the first processing, the calculator calculates a first score based on a state of pixels of the character. In the second processing, the calculator calculates a second score based on a match ratio between the pixels and the extracted pixels. In the third processing, the calculator calculates a third score based on a ratio of a length of the character image in first and second direction. In the fourth processing, the calculator calculates a fourth score based on a comparison result between the detected result and preset patterns. The calculator calculates a certainty of the reading by using one of the first, second, third, or fourth score.