Abstract: Methods and apparatus are provided for determining quantization parameter predictors from a plurality of neighboring quantization parameters. An apparatus includes an encoder for encoding image data for at least a portion of a picture using a quantization parameter predictor for a current quantization parameter to be applied to the image data. The quantization parameter predictor is determined using multiple quantization parameters from previously coded neighboring portions. A difference between the current quantization parameter and the quantization parameter predictor is encoded for signaling to a corresponding decoder.

Abstract: A computer system obtains, in electronic format, a training dataset. The training dataset comprises a plurality of training images from a plurality of agricultural plots. Each training image is from a respective agricultural plot in the plurality of agricultural plots and comprises at least one identified fruit. The computer system determines, for each respective fruit in each respective training image in the plurality of training images, a corresponding contour. The computer system trains an untrained or partially trained computational model using at least the corresponding contour for each respective fruit in each respective training image in the plurality of training images, thereby obtaining a first trained computational model that is configured to identify fruit in agricultural plot images.

Abstract: Described herein is a novel method and system for segmentation of the spine using 3D volumetric data. In embodiments, a method includes an extracting step, localization step, and segmentation step. The extracting step comprises detecting the spine centerline and the spine canal centerline. The localization step comprises localizing the vertebra and intervertebral disc centers. Background and foreground constraints are created for each vertebra digit. Segmentation is performed for each vertebra digit and based on the hard constraints.

Abstract: In one embodiment, a method includes receiving an image associated with an object in an environment, the image being captured by sensors associated with a vehicle, generating a feature representation of the image, determining a potential ground control point associated with the object based on the feature representation of the image, determining a predetermined location reading based on the potential ground control point, calculating a differential relative to the predetermined location reading based on the potential ground control point, and determining a location of the vehicle based on the differential and the predetermined location reading based on the potential ground control point.

Abstract: A method for controlling a mobile communication device using a user's distance specifying algorithm, the method comprises a first process of receiving an input image from a camera to generate first data; a second process of determining whether there is a candidate group that can be determined as a face using the first data and second data including a user's distance information; a third process of extracting, if there is a candidate group that can be determined as a face, from a plurality of third data including location information of the candidate group, fourth data having a minimum distance between a center of the first data and a center of the third data; and a fourth process of comparing the fourth data with a predetermined user's distance to determine whether it has a value within a tolerance range.

Abstract: A method for searching a video and equipment with a video search function are provided. The method for searching a video includes constructing a video DB by analyzing continuity of a tag given to an appearing object and extracting section information about the tag, and detecting video information. An object may be recognized, a video database may be constructed, and a video may be searched on the basis of analysis based on an artificial intelligence (AI) model through a 5G network.

Abstract: A method and system is provided for determining the denomination and related data for a currency item using a personal computing device, such as a mobile phone. The device includes or is connected to an image capture device that is preferably a digital video camera. At least one image of a target currency item is captured then processed for image quality. A further processing of the image includes a coordinate mapping. A comparison is made between individual pixels of the processed image based on the assigned coordinate mapping with a database of reference currency images to determine the currency denomination. Additional processing of the currency image provides the date and other data regarding the target currency item. A market value for the target currency item is identified by reference to a valuation database using the data determined for the currency item.

Abstract: A method for detecting a parallax problem in sensor data from two sensors spaced apart from each other and at least partly capturing the same environment, wherein at least one of the sensors provides distance information. The method includes obtaining the acquired sensor data from the sensors; assigning measured values in the acquired sensor data of one sensor to corresponding measured values in the acquired sensor data of the other sensor, wherein the assignment takes the respective imaging conditions of the two sensors into account; consecutively numbering the measured values in the sensor data; checking whether a sorting order of the numbering of the measured values that correspond to each other matches, a parallax problem being determined in response to a sorting order not matching; and outputting a test result. Also disclosed is an apparatus for detecting a parallax problem in sensor data from two sensors and a transportation vehicle.

Abstract: There is provided a filing device, a filing method, and a program that cause a still image and a motion image to be associated with each other, which are captured without being associated with each other. The problem is solved through a filing device including an acquisition unit that acquires at least one still image and at least one motion image, a still image analysis unit that analyzes a pixel value of the still image and extracts still image information, a motion image analysis unit that analyzes a pixel value of the motion image and extracts motion image information, and an associating unit that associates the still image with the motion image by comparing the still image information with the motion image information.

Abstract: Systems and methods are provided for suggesting reminders from content displayed on a mobile device. An example method may include analyzing content generated by a first mobile application and displayed on a display of a mobile device, and determining that the content suggests an event, the event including at least one entity. The method may also include providing an assistance window requesting confirmation for adding a reminder for the event in a second mobile application responsive to determining that the content suggests the event, and adding the reminder via the second mobile application responsive to receiving the confirmation. In some implementations the first mobile application is a messaging application.

Abstract: A method for estimating text heights of text line images includes estimating a text height with a sequence recognizer. The method further includes normalizing a vertical dimension and/or position of text within a text line image based on the text height. The method may also further include calculating a feature of the text line image. In some examples, the sequence recognizer estimates the text height with a machine learning model.

Abstract: An apparatus including an interface and a processor. The interface may be configured to receive pixel data. The processor may be configured to generate a plurality of video frames in response to the pixel data received from the interface, perform computer vision operations to detect objects in the video frames, extract features of the objects in response to characteristics of the objects determined using the computer vision operations, identify a person in the video frames based on the features, detect an event based on the person identified and generate a notification in response to detecting the event and a permission status. The permission status may suppress sending the notification when the permission status for the person identified corresponds to denying the notification and enable sending the notification when the permission status does not correspond to denying the notification.

Abstract: A local extension method for segmentation of anatomical treelike structures includes receiving an initial segmentation of 3D image data including an initial treelike structure. A target point in the 3D image data is defined, and a region of interest based on the target point is extracted to create a sub-image. Highly tubular voxels are detected in the sub-image, and a spillage-constrained region growing is performed using the highly tubular voxels as seed points. Connected components are extracted from the results of the region growing. The extracted components are pruned to discard components not likely to be connected to the initial treelike structure, keeping only candidate components likely to be a valid sub-tree of the initial treelike structure. The candidate components are connected to the initial treelike structure, thereby extending the initial segmentation in the region of interest.

Abstract: Aspects of the disclosure provide a method for denoising an image. The method can include receiving an acquired image from an image acquisition system, and processing the acquired image with a nonlinear diffusion coefficient based filter having a diffusion coefficient that is calculated using gradient vector orientation information in the acquired image.

Abstract: An apparatus accesses label data and training images corresponding to a geographic area; and provides the label data and training images to a training model. The training model comprises of at least a predictor model and an alignment model. The predictor model is configured to receive an image and provide a prediction corresponding to the image. The alignment model is configured to generate a transformed prediction based on aligning the label data and the prediction. The apparatus executes a loss engine to iteratively receive the label data and the transformed prediction, evaluate a loss function based on the label data and the transformed prediction, and cause weights of the predictor model and the alignment model to be updated based on the evaluated loss function to cause the predictor and alignment models to be trained.

Abstract: A statistical learning technique that does not rely upon paired imaging information is described herein. The technique may be computer-implemented and may be used in order to train a statistical learning model to perform image synthesis, such as in support of radiation therapy treatment planning. In an example, a trained statistical learning model may include a convolutional neural network established as a generator convolutional network, and the generator may be trained at least in part using a separate convolutional neural network established as a discriminator convolutional network. The generator convolutional network and the discriminator convolutional network may form an adversarial network architecture for use during training. After training, the generator convolutional network may be provided for use in synthesis of images, such as to receive imaging data corresponding to a first imaging modality type, and to synthesize imaging data corresponding to a different, second imaging modality type.

Abstract: Methods and apparatus to capture images are disclosed. An example apparatus disclosed herein includes an image sensor to observe a first image of a face, and a logic circuit to determine whether the face in the first image is in a first orientation with respect to the image sensor, and when the face in the first image is in the first orientation: trigger the image sensor to capture a second image of the face, and perform a facial recognition process on the second image, but not on the first image.

Abstract: Devices and techniques are generally described for computer vision techniques for safety hazard detection. A frame of image data representing a physical environment is received. In some examples, a first object represented in the frame of image data may be detected. A determination may be made that the first object is of a first class. A first zone represented in the frame of image data may be identified. The first zone may correspond to a ground surface of the physical environment. A determination may be made that the first object at least partially overlaps with the first zone. A first rule associated with the first zone may be determined. The first rule may restrict objects of the first class from being present within the first zone. Output data may be generated indicating that the first object is at least partially within the first zone, in violation of the first rule.

Abstract: In one example embodiment, a computer-implemented method for extracting information from imagery includes obtaining data representing a sequence of images, at least one of the sequence of images depicting an object. The method includes inputting the sequence of images into a machine-learned information extraction model that is trained to extract location information from the sequence of images. The method includes obtaining as an output of the information extraction model in response to inputting the sequence of images, data representing a real-world location associated with the object depicted in the sequence of images.

Abstract: Provided is a method, system, and computer program product for storing images across multiple distributed computing systems according to image sensitivity. The method comprises identifying an image and analyzing the image to identify sensitive information in the image. The method further comprises splitting the image into a sensitive portion and a non-sensitive portion. The method further comprises storing the sensitive portion of the image in a first distributed computing system and storing the non-sensitive portion of the image in a second distributed computing system.