Abstract: The present disclosure is directed to apparatuses, systems and methods for automatically classifying images of occupants inside a vehicle. More particularly, the present disclosure is directed to apparatuses, systems and methods for automatically classifying images of occupants inside a vehicle by comparing current image feature data to previously classified image features.

Abstract: A number-of-occupants detection system (1) is configured to include an image acquisition unit (110) that acquires, from an imaging apparatus mounted on a first vehicle, an image of a second vehicle existing around the first vehicle and a count unit (120) that counts the number of occupants of the second vehicle using the image acquired by the image acquisition unit (110).

Abstract: An authentication device includes an illumination device, a camera, and a processor configured to obtain, by capturing by the camera, an image including a subject illuminated with light of the illumination device, extract a target region from the obtained image, and perform determination that processing related to authentication is not allowed when the target region includes a first part of a first luminance value having a difference not less than a threshold value from a second luminance value of a second part of the target region.

Abstract: Recent progress of neural networks has shown significant progress over human pose estimation tasks. Pose estimation could be categorized into monocular 2D pose estimation, multi-view 3D pose estimation, and single view 3D pose estimation, where recently 3D pose is getting more attention to be applied to AR/VR, game and human computer interaction applications. However, current academic benchmarks on human 3D pose estimation only respect performance on their relative pose. The root positioning over the time, in another word, the “trajectory” of the whole body in 3D space is not considered well enough. Applications such as motion capture, not only require precise relative pose of the body but also the root position of the whole body in 3D space. Therefore, an efficient monocular full 3D pose recovery model from 2D pose input is described herein, which is able to be applied to the above applications.

Abstract: The present disclosure is directed to apparatuses, systems and methods for automatically classifying digital images of occupants inside a vehicle. More particularly, the present disclosure is directed to apparatuses, systems and methods for automatically classifying digital images of occupants inside a vehicle by comparing current image data to previously classified image data.

Abstract: The present disclosure is directed to apparatuses, systems and methods for automatically classifying digital images of occupants inside a vehicle. More particularly, the present disclosure is directed to apparatuses, systems and methods for automatically classifying digital images of occupants inside a vehicle by comparing current image data to previously classified image data.

Abstract: The present disclosure provides an age recognition method, a non-transitory computer-readable storage medium, and an electronic device. The method includes extracting feature points of a face in an image, preprocessing the image to extract global features of the face, extracting local features of the face based on the feature points, determining an age feature of the face according to the global features and the local features, and inputting the age feature into a pre-trained age recognition model to obtain an age value corresponding to the face in the image.

Abstract: A server device includes: a data collection unit that collects vehicle behavior data; a scene extraction unit that extracts, from the collected vehicle behavior data, driving scenes corresponding to the behavior of the vehicle and a scene feature amount of each of the driving scenes; an abnormality detection unit that calculates a degree of abnormality which represents an extent of deviation of the scene feature amount of each of the extracted driving scenes relative to a driving model, and detects a driving scene including a location of an abnormality using the calculated degree of abnormality; a section determination unit that extracts driving scenes satisfying a predetermined condition from among the detected driving scene and a plurality of driving scenes continuous to the driving scene, and determines, as an abnormal behavior section, a time range defined by the total continuation time of the extracted driving scenes; an image request unit that requests, from the vehicle, one or more captured images acco

Abstract: A mobile dimensioning device, i.e. a mobile dimensioner, is described that uses a dynamic accuracy while still being compatible with the NIST standard. Even if the accuracy division is dynamic and not predetermined, a mobile dimensioning device of the present invention reports the actual dimensioning prior to measurement capture and can therefore be certified and used in commercial transactions.

Abstract: Systems and methods for identifying semantically and/or visually related information among a set of content items, such content items that include similar concepts or that have similar visual aspects, are disclosed. The disclosed techniques provide tools for identifying related information among various content items, such as text pages and documents, presentation slides and slide decks, etc. The disclosed techniques provide improved methods for searching among content items, organizing content items into categories, and pruning redundant content. Furthermore, the disclosed techniques provide improvements to computation of various metrics, including usage, performance, and impact metrics.

Abstract: One variation of a method for managing parking within a parking structure includes: after detecting a license plate number on a vehicle in an image captured by an entry camera unit arranged near an entry of the parking structure, labeling the license plate as unverified in a data log associated with the parking structure; in response to absence of validation of the license plate within a grace period after entry of the vehicle, relabeling the license plate as in-violation in the data log; in response to detecting the license plate in an image captured by a mobile enforcement unit moving through the parking structure, querying the data log for a status of the license plate; and, in response to the license plate labeled as in-violation in the data log, serving a prompt to an enforcement personnel to issue a citation to a vehicle carrying the license plate.

Abstract: Systems and methods designed to detect a human being falling as opposed to an inanimate object. Generally, the systems and methods will utilize a depth camera, which will often image in the NIR spectrum to detect a falling object. The portion detected as a falling object will often be detected as separating from a point cloud indicative of one object in contact with another. Should such a separation be detected, the systems and methods will utilize a thermal sensor, often a camera imaging in the LWIR spectrum, to determine if the falling portion has a heat signature indicative of a human being.

Abstract: Methods are provided for operating a medical X-ray device to improve the image quality of an X-ray examination. In one example, the method includes recording at least one first X-ray image of a body region as a mask image; providing a first subsequent image and recording a second X-ray image of the body region, wherein the second X-ray image represents the body region at a later recording time than the first subsequent image; determining a degree of deviation relating to a deviation between the first subsequent image and the second X-ray image; determining an averaging amount in dependence on the degree of deviation; generating a second subsequent image from the second X-ray image or from the first subsequent image together with the second X-ray image, wherein the averaging amount specifies the proportions in which the first subsequent image and the second X-ray image are mixed; and forming an overall image from the mask image and the second subsequent image.

Abstract: Representative embodiments disclose machine learning classifiers used in scenarios such as speech recognition, image captioning, machine translation, or other sequence-to-sequence embodiments. The machine learning classifiers have a plurality of time layers, each layer having a time processing block and a depth processing block. The time processing block is a recurrent neural network such as a Long Short Term Memory (LSTM) network. The depth processing blocks can be an LSTM network, a gated Deep Neural Network (DNN) or a maxout DNN. The depth processing blocks account for the hidden states of each time layer and uses summarized layer information for final input signal feature classification. An attention layer can also be used between the top depth processing block and the output layer.

Abstract: Embodiments describing an approach to evaluate text and image consistency. Receiving one or more images. Receiving one or more text documents. Identifying relevant text in the one or more text documents. Determining the consistency between the one or more images and the one or more text documents. Creating one or more image and text consistency scores based on the determined consistency between the one or more images and the one or more text documents, and outputting the one or more image and text consistency scores for evaluating text and image consistency.

Abstract: Monitoring target matching means 71 matches monitoring targets shown in video captured by one or more imaging devices, and identifies monitoring targets estimated to be the same monitoring target, as an identified monitoring target. Target object identifying means 72 identifies a desired target object from one or more identified monitoring targets captured, using imaging times of each of the one or more identified monitoring targets.

Abstract: Disclosures herein teach applying a set of sections spanning a down-sampled version of an image of a road-scene to a low-fidelity classifier to determine a set of candidate sections for depicting one or more objects in a set of classes. The set of candidate sections of the down-sampled version may be mapped to a set of potential sectors in a high-fidelity version of the image. A high-fidelity classifier may be used to vet the set of potential sectors, determining the presence of one or more objects from the set of classes. The low-fidelity classifier may include a first Convolution Neural Network (CNN) trained on a first training set of down-sampled versions of cropped images of objects in the set of classes. Similarly, the high-fidelity classifier may include a second CNN trained on a second training set of high-fidelity versions of cropped images of objects in the set of classes.

Abstract: Systems, methods, devices and computer readable media for determining a geographical location of a license plate are described herein. A first image of a license plate is acquired by a first image acquisition device of a camera unit and a second image of the license plate is acquired by a second image acquisition device of the camera unit. A three-dimensional position of the license plate relative to the camera unit is determined based on stereoscopic image processing of the first image and the second image. A geographical location of the camera unit is obtained. A geographical location of the license plate is determined from the three-dimensional position of the license plate relative to the camera unit and the geographical location of the camera unit. Other systems, methods, devices and computer readable media for detecting a license plate and identifying a license plate are described herein.

Abstract: A computer implemented method of fingertip centroid identification in real-time, implemented on a computer system comprising a processor, memory, and a camera system. The processor receives image data from the camera system; runs a first kernel comprising a set of concentric closed shapes over image data to identify an occupancy pattern in which internal closed shapes are at least nearly fully occupied, and in which a subsequent closed shape has at least a relatively low occupancy level, so as to identify one or more fingertips in the image data; for each identified fingertip, runs a second kernel over the identified one or more fingertips to establish a best fit closed shape which covers each identified fingertip; calculates a centroid for each best fit closed shape which corresponds to an identified fingertip; and stores in the memory the calculated centroids for the identified one or more fingertips.

Abstract: The invention relates to a method for capturing an object (9) in an environmental region (8) of a motor vehicle (1) based on a sequence of images (10, 11) of the environmental region (8), which are provided by means of a camera (4) of the motor vehicle (1), including the steps of: recognizing a first object feature (24) in a first image (10) of the sequence, wherein the first object feature (24) describes at least a part of the object (9) in the environmental region (8), estimating a position of the object (9) in the environmental region (8) based on a predetermined movement model, which describes a movement of the object (9) in the environmental region (8), determining a prediction feature (26) in a second image (11) following the first image (10) in the sequence based on the first object feature (24) and based on the estimated position, determining a second object feature (25) in the second image (11), associating the second object feature (25) with the prediction feature (26) in the second image (11) if a