Abstract: Advanced driver assistance (ADAS) systems and methods for a vehicle comprise capturing an image using a monocular camera system of the vehicle and detecting a landmark in the image using a deep neural network (DNN) trained with labeled training data including generating a bounding box for the detected landmark, predicting a depth of pixels in the image using a convolutional neural network (CNN) trained with unlabeled training data captured by a stereo camera system, filtering noise by averaging predicted pixel depths for pixels in the region of the bounding box to obtain an average depth value for the detected landmark, determining a coordinate position of the detected landmark using its average depth value, and performing at least one ADAS feature using the determined coordinate position of the detected landmark.

Abstract: In an image processing apparatus, a face detection unit detects a face area from an input image, and an image processing unit extracts a high luminance area from the face area to generate a correction amount of the high luminance area and corrects the high luminance area using the correction amount. Then, in a case where a plurality of face areas is detected, the image processing unit adjusts the correction amount based on feature information of the plurality of face areas.

Abstract: The present disclosure relates to an image correction method and device. The method includes detecting an input image to obtain boundary straight lines. The method includes determining vertices based on the obtained boundary straight line. The method includes determining an estimated height-to-width ratio based on the obtained vertices. The method includes performing perspective transformation on the input image based on the estimated height-to-width ratio.

Abstract: In a magnetic resonance fingerprinting method and apparatus for improved determination of local parameter values of an examination object, in which at least two signal comparisons of acquired picture element time series are carried out with comparison signal curves for determination of parameter values. A further (subsequent) signal comparison takes into account results of a preceding signal comparison. This multi-stage determination of parameter values allows an increase of the spatial resolution and the precision with which the parameter values can be determined.

Abstract: An image processing apparatus is provided. The image processing apparatus according to an exemplary embodiment includes a communicator configured to receive an image, and a processor configured to generate a first image obtained by performing image processing on the received image by using a parameter for image processing, generate a second image obtained by reducing the first image at a predetermined ratio, and extract respective visual features from the first image and the second image, wherein the processor is further configured to adjust the parameter to allow a difference between the visual feature of the first image and the visual feature of the second image to be within a predetermined range.

Abstract: Disclosed herein are embodiments of systems, methods, and products for a webly supervised training of a convolutional neural network (CNN) to predict emotion in images. A computer may query one or more image repositories using search keywords generated based on the tertiary emotion classes of Parrott's emotion wheel. The computer may filter images received in response to the query to generate a weakly labeled training dataset labels associated with the images that are noisy or wrong may be cleaned prior to training of the CNN. The computer may iteratively train the CNN leveraging the hierarchy of emotion classes by increasing the complexity of the labels (tags) for each iteration. Such curriculum guided training may generate a trained CNN that is more accurate than the conventionally trained neural networks.

Abstract: Provided are a proposals processing method and related products for object detection, where the method comprising: determining a score threshold according to a statistics property of scores of a first quantity of proposal; selecting a second quantity of proposals with a score higher than the score threshold from the first quantity of proposals; and suppressing to the second quantity of proposals to get a third quantity of non-overlapping proposals with high scores, as such, time complexity for selecting a second quantity of proposals is reduced, and efficiency for selection of the proposals to be used for suppression is improved, thereby improving proposal processing efficiency and object detection efficiency.

Abstract: Processing an image includes acquiring, by the image processing apparatus, a target image, extracting a shape of a target object included in the target image, determining a category including the target object based on the extracted shape, and storing the target image by mapping the target image with additional information including at least one keyword related to the category.

Abstract: A facial recognition method using online sparse learning includes initializing target position and scale, extracting positive and negative samples, and extracting high-dimensional Haar-like features. A sparse coding function can be used to determine sparse Haar-like features and form a sparse feature matrix, and the sparse feature matrix in turn is used to classify targets.

Abstract: Systems, methods and computer program products for image recognition in which instructions are executable by a processor to dynamically generate simulated documents and corresponding images, which are then used to train a fully convolutional neural network. A plurality of document components are provided, and the processor selects subsets of the document components. The document components in each subset are used to dynamically generate a corresponding simulated document and a simulated document image. The convolutional neural network processes the simulated document image to produce a recognition output. Information corresponding to the document components from which the image was generated is used as an expected output. The recognition output and expected output are compared, and weights of the convolutional neural network are adjusted based on the differences between them.

Abstract: An apparatus may include an ultrasonic sensor array, a light source system and a control system. Some implementations may include an ultrasonic transmitter. The control system may be operatively configured to control the light source system to emit light that induces acoustic wave emissions inside a target object. The control system may be operatively configured to select a first acquisition time delay for the reception of acoustic wave emissions primarily from a first depth inside the target object. The control system may be operatively configured to acquire first ultrasonic image data from the acoustic wave emissions received by the ultrasonic sensor array during a first acquisition time window. The first acquisition time window may be initiated at an end time of the first acquisition time delay.

Abstract: A reconstruction processing method. The method includes an image updating step (Step S2) of updating a reconstruction image by an iterative approximation method, and a weighting coefficient map updating step (Step S4). In the weighting coefficient map updating step, a weighting coefficient map relative to prior knowledge is generated from the reconstruction image obtained by updating an image in the image updating step (Step S2), and a weighting coefficient of the prior knowledge relative to each pixel is controlled in accordance with the weighting coefficient map, whereby a weighting coefficient map is updated. As described above, the weighting coefficient map relative to the prior knowledge is generated from the reconstruction image (during estimation) obtained by updating the image, and the weighting coefficient of the prior knowledge relative to each pixel is controlled in accordance with the weighting coefficient map.

Abstract: The present disclosure relates to an image processing method and device, a computer readable storage medium, and an electronic device. The method includes: acquiring an image to be processed; detecting a foreground target of the image, and recognizing the foreground target, to obtain a foreground classification result; detecting a background area of the image, and recognizing the background area, to obtain a background classification result; and obtaining a classification label of the image based on the foreground classification result and the background classification result.

Abstract: The present disclosure provides devices, systems and methods for mapping of traffic boundaries.

Abstract: Systems and methods for object recognition and association with an identity are disclosed. Systems and methods for object recognition and association with an identity are disclosed. In one embodiment, in an information processing apparatus comprising at least one computer processor, a method for object recognition and association with an identity may include: (1) receiving, from a first image capture device at a facility, a first image or a video; (2) recognizing, in the first image or video, an individual having a physical characteristic and an object in proximity to the individual; (3) associating the physical characteristic and the object with the individual; and (4) storing the association in a database.

Abstract: Some implementations can include a computer-implemented method and/or system for automatic suggestion to share images. The method can include identifying a plurality of images associated with a user and detecting one or more entities in the plurality of images. The method can also include constructing an aggregate feature vector for the plurality of images based on the one or more entities in the plurality of images and determining that the aggregate feature vector matches a first cluster. The method can further include, in response to determining that the aggregate feature vector matches the first cluster, providing a suggestion to the user for an image composition based on the plurality of images.

Abstract: An image-processing method, an apparatus and a device are provided. The method includes: controlling a primary camera to capture a plurality of primary images, and simultaneously controlling a secondary camera to capture a plurality of secondary images; obtaining a reference primary image from the primary images, and obtaining a reference secondary image co-captured with the reference primary image from the secondary images; starting a multithread-based parallel processing mechanism, calling a first thread to perform a multiframe-synthesizing noise-reduction process for the primary images so as to generate a target primary image, and simultaneously calling a second thread to obtain depth information according to the reference primary image and the reference primary image; and performing a blurring process for a background region of the target primary image according to the depth information. Thus, it improves the accuracy of calculating the depth information and the image-processing efficiency thereof.

Abstract: Provided are an image detection device, an image detection method and a program, which are capable of improving correspondence to a target deformation by optimizing a template shape, when performing target detection using template matching. An image detection device 100 for detecting a target from an input image comprises: a template generation unit 10 that generates a template for detecting a target; a mask generation unit 20 that generates a mask which shields a portion of the template, on the basis of temporal variations of a feature point extracted from an area including the image target; and a detection unit 30 that detects the target from the image using the template a portion of which is shielded by the mask.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for tracking objects over a sequence of images. In one aspect, a process includes receiving data specifying visual attributes of a particular object detected in an image. Data specifying visual attributes of a tracked object is identified. A level of similarity is determined between the particular object and the tracked object based on comparison of visual attributes of the particular object with corresponding visual attributes of the tracked object. A first normalization function is used to normalize a level of similarity between first corresponding visual attributes of the particular object and the tracked object. A second normalization function is used to normalize a level of similarity between second corresponding visual attributes of the particular object and the tracked object. A determination is made, based on the level of similarity, whether the particular object matches the tracked object.

Abstract: Disclosed are techniques for detecting presence of shadowing introduced in a system. The techniques entail geometrically projecting, along a virtual or actual propagation direction of laser beam scanning, points of a lower convex hull (CH) planar object in a direction toward a higher CH planar object to establish projection locations of the points of the lower CH planar object along the propagation direction. Detection of whether the projection locations are in proximity to locations of points of the higher CH planar object is performed. In response to detecting locations in proximity for each source laser beam scanning, a shadowing event indication is generated to alert a user of the system that shadowing has been introduced in the system.