Abstract: A system and method for making categorized music tracks available to end user applications. The tracks may be categorized based on computer-derived rhythm, texture and pitch (RTP) scores for tracks derived from high-level acoustic attributes, which is based on low level data extracted from the tracks. RTP scores are stored in a universal database common to all of the music publishers so that the same track, once RTP scored, does not need to be re-RTP scored by other music publishers. End user applications access an API server to import collections of tracks published by publishers, to create playlists and initiate music streaming. Each end user application is sponsored by a single music publisher so that only tracks capable of being streamed by the music publisher are available to the sponsored end user application.

Abstract: In various examples, surface profile estimation and bump detection may be performed based on a three-dimensional (3D) point cloud. The 3D point cloud may be filtered in view of a portion of an environment including drivable free-space, and within a threshold height to factor out other objects or obstacles other than a driving surface and protuberances thereon. The 3D point cloud may be analyzed—e.g., using a sliding window of bounding shapes along a longitudinal or other heading direction—to determine one-dimensional (1D) signal profiles corresponding to heights along the driving surface. The profile itself may be used by a vehicle—e.g., an autonomous or semi-autonomous vehicle—to help in navigating the environment, and/or the profile may be used to detect bumps, humps, and/or other protuberances along the driving surface, in addition to a location, orientation, and geometry thereof.

Abstract: A data item is identified on a device. A neural network that includes an adversarial transformation subnetwork is applied to the data item to generate a modified data item. Output indicative of the modified data item is caused to be presented on the device. The neural network further comprises an encoder and a decoder. The neural network is trained in at least two stages. At least one of the encoder and the decoder is trained in a first stage and the adversarial transformation subnetwork is trained in a second stage.

Abstract: An information processing device of the present invention includes a detection means that detects the content of an image, a determination means that determines a processing mode for the image based on the result of detection of the content of the image, and an execution means that executes processing for a captured image corresponding to the processing mode.

Abstract: According to some embodiments, a method includes accessing a video generated by a camera in a physical environment. The method further includes identifying, from a first video frame of the video, a first object of interest corresponding to a physical object in the physical environment. The method further includes storing a record for the first object of interest that includes a unique identifier. The method further includes identifying a second object of interest from a second video frame. The method further includes comparing variables of the first object of interest to variables of the second object of interest. The method further includes determining that the variables of the first object of interest match the variables of the second object of interest and then assigning the unique identifier to the second object of interest. The method further includes updating the record to include the second object of interest.

Abstract: An apparatus to facilitate partially-frozen neural networks for efficient computer vision systems is disclosed. The apparatus includes a frozen core to store fixed weights of a machine learning model, one or more trainable cores coupled to the frozen core, the one or more trainable cores comprising multipliers for trainable weights of the machine learning model, and wherein the alpha blending layer includes a trainable alpha blending parameter, and wherein the trainable alpha blending parameter is a function of a trainable parameter, a sigmoid function, and outputs of frozen and trainable blocks in a preceding layer of the machine learning model.

Abstract: Various systems and methods for for clustering facial images in, for example, surveillance systems.

Abstract: There is provided a method of determining a three-dimensional shape of an insert for insertion into an ear. The method includes receiving image data corresponding to a two-dimensional image of an ear, processing the image data to measure at least one biometric feature of the ear, the at least one biometric feature being indicative of a three-dimensional shape of at least part of the ear, and determining a three-dimensional shape of an insert for insertion into the ear by matching said at least one biometric feature with one of a plurality of pre-stored three-dimensional shapes. Each pre-stored three-dimensional shape corresponds to a respective ear.

Abstract: A method of task-adaptive pre-processing (TAPP) for neural image compression is performed by at least one processor and includes generating a substitutional image, based on an input image, using a TAPP neural network, and encoding the generated substitutional image to generate a compressed representation, using a first neural network.

Abstract: A returned goods processing system may include a camera configured to acquire human pose image data of a given user while processing a returned item. The system may also include a returned goods processing server configured to store reference human pose image data of sample users while processing returned items. The stored human pose image data may be indicative of compliance with an acceptable return processing procedure. The returned good processing server may also be configured to compare the acquired human pose image data with the stored reference human pose image data to determine compliance with the acceptable return processing procedure, and generate a notification based upon the comparing. The returned goods processing server may also be configured to perform machine learning to update the stored reference human pose image data based upon the acquired human pose image data.

Abstract: An apparatus to facilitate partially-frozen neural networks for efficient computer vision systems is disclosed. The apparatus includes a frozen core to store fixed weights of a machine learning model, one or more trainable cores coupled to the frozen core, the one or more trainable cores comprising multipliers for trainable weights of the machine learning model, and wherein the alpha blending layer includes a trainable alpha blending parameter, and wherein the trainable alpha blending parameter is a function of a trainable parameter, a sigmoid function, and outputs of frozen and trainable blocks in a preceding layer of the machine learning model.

Abstract: The invention relates to a monitoring device (10) for recognizing persons in a monitoring region (2), the monitoring region (2) being video-monitored by means of at least one camera (6) and the camera (6) being designed to provide monitoring images (7) to the monitoring device (10) as video data, the monitoring device comprising: —a feature determination apparatus (13), the feature determination apparatus (13) being designed to determine a feature vector (19) for each object in at least one of the monitoring images (7); —a person recognition apparatus (16), the person recognition apparatus (16) being designed to detect in the monitoring images (7) a person to be recognized (11), on the basis of the determined feature vector and/or the determined feature vectors (19) of the feature determination apparatus (13) and/or a combined feature vector (18); —an association apparatus (14), the association apparatus (14) being designed to determine a feature vector (19) for each person to be recognized (11) and each asso

Abstract: An object recognition method uses a sensor configured to acquire a position of an object existing in a surrounding environment as point clouds including a plurality of detection points in a top view. The method includes: grouping point clouds according to a proximity; determining, when performing polygon approximation on the grouped point clouds, whether at least part of the detection points constituting the grouped point clouds are located in a blind zone of an approximate polygon acquired by the polygon approximation with respect to the sensor; recognizing the grouped point clouds as point clouds corresponding to plural objects when it is determined that the detection points are located in the blind zone with respect to the sensor; and recognizing the grouped point clouds as point clouds corresponding to a single object of the approximate polygon when it is determined that the detection points are not located in the blind zone.

Abstract: An image correction device including: an image data acquisition part for acquiring the image data obtained by a camera; and a correction part for correcting the image data on the basis of a first difference that is a difference between a first image height obtained by light having entered the camera at a first incident angle and a first reference image height obtained by light having entered the camera at the first incident angle when the camera is free of manufacturing errors.

Abstract: A bi-directional spatial-temporal transformer neural network (BDSTT) is trained to predict original coordinates of a skeletal joint in a specific frame through relative relationships of the skeletal joint to other joints and to the state of the skeletal joint in other frames. Obtain a plurality of frames comprising coordinates of the skeletal joint and coordinates of other joints. Produce a spatially masked frame by masking the original coordinates of the skeletal joint. Provide the specific frame, the spatially masked frame, and at least one more frame to a coordinate prediction head of the BDSTT. Obtain, from the coordinate prediction head, a prediction of coordinates for the skeletal joint. Adjust parameters of the BDSTT until a mean-squared error, between the prediction of coordinates for the skeletal joint and the original coordinates of the skeletal joint, converges.

Abstract: The present disclosure relates to systems, methods, non-transitory computer-readable media, and apparatuses relating to a camera, an output device, a locomotion system, and a processing circuit including a processor and a memory, the processing circuit is configured to move a mobile platform system using the locomotion system in a cruise mode within an area, determine, using a camera of the mobile platform system, image or video data of a plurality of individuals within the area while moving in the cruise mode, select a first target individual and a second target individual of the plurality of individuals based on analyzing the image or video data, determine first contextual information of the first target individual and second contextual information of the second target individual based on analyzing the image or video data, determine first characteristic of the first target individual based on the first contextual information and second characteristic of the second target individual based on the second conte

Abstract: The field of the invention relates to computer vision systems and methods providing real time data analytics on detected people or objects in the home environment or other environments. It is based on an embedded engine that analyses an image from a raw sensor and virtualised the image into a digital representation enabling a digital understanding of the environment while guarantying privacy. It comprises multiple image processing blocks and embedded firmware.

Abstract: An information processing apparatus includes: an acquisition unit that acquires a two-dimensional image of a person, and a registered image group including two-dimensional registered images and three-dimensional registered images of a plurality of registrants; and a selection unit that selects a type of a registered image to be used for matching from one of the two-dimensional registered images and the three-dimensional registered images based on the two-dimensional image, prior to the matching the two-dimensional image with the registered image group.

Abstract: An imaging system can receive an image of a portion of an environment. The environment can include an object, such as a hand or a display. The imaging device can identify a data stream from an external device, for instance by detecting the data stream in the image or by receiving the data stream wirelessly from the external device. The imaging device can detect a condition based on the image and/or the data stream, for instance by detecting that the object is missing from the image, by detecting that a low resource at the imaging device, and/or by detecting visual media content displayed by a display in the image. Upon detecting the condition, imaging device automatically determines a location of the object (or a portion thereof) using the data stream and/or the image. The imaging device generates and/or outputs content that is based on the location of the object.

Abstract: An automatic video production device that performs fitting or fusion processing of a predetermined child video object that is not included in a parent video into a predetermined video scene of the parent video content as a video fusion source, the automatic video production device including a video content interpreting unit that interprets the entirety or a part of the video content of the parent video content, or divides the parent video content into a predetermined parent video scene to generate processing data that is predetermined parent video interpretation data obtained by converting a content interpretation result of the parent video or a division result of the parent video scene into data in a predetermined description format.