Abstract: Systems, video cameras, and methods for a peer-to-peer surveillance camera architecture using a shared metadata datastore are described. Each video camera may be configured to detect objects of interest in the video data captured by their respective image sensors and selectively upload metadata from object detection events to the shared metadata datastore. Responsive to detecting an object of interest, each camera may also use the metadata from the object detection to query the datastore for similar events detected by other video cameras. Responsive to the query, the video camera may receive a peer camera identifier and initiate peer-to-peer communication with the identified peer video camera to exchange video metadata directly with that peer video camera.

Abstract: An image processing apparatus includes a detector configured to detect at least one object included in at least one of visible light image data and non-visible light image data that are generated by imaging, and a processor configured to determine a tracking object from the at least one object, to generate combined image data by combining the visible light image data and the non-visible light image data, and to track the tracking object by using the combined image data. The processor combines the visible light image data and the non-visible light image data in accordance with information related to the at lease one object detected by the detector.

Abstract: A panoramic surveillance video generating apparatus includes: a plurality of image sensors configured to acquire a plurality of surveillance videos; and at least one processor configured to: generate a panoramic video by combining the plurality of surveillance videos, detect an event occurring in the panoramic video, set a partial region of the panoramic video as a surveillance region, output a video of the surveillance region to a virtual channel, and adjust at least one of a quantity of the surveillance region, a size of the surveillance region, and a position of the surveillance region based on an event detection result.

Abstract: Apparatuses, systems, and techniques are presented to select segments of content. In at least one embodiment, one or more neural networks are used to select one or more video segments from one or more video files based at least in part upon an indication of one or more emotions, of one or more viewers of the one or more video segments, during presentation of the one or more video segments.

Abstract: A welding information providing device is proposed. The device may include a body provided to cover at least a portion of a worker' face and including a first region facing the worker' face and a second region located in a region opposite to the first region. The device may also include a camera unit located in the second region of the body and capturing a welding video to acquire a video frame. The device may further include a display unit located in the first region of the body and providing a video generated through the video frame.

Abstract: A method of setting up non-randomly positioned markers in a field of view of a camera bar, and tracking a pose of the camera bar, including: (i) a camera bar of a video camera viewing a field of view; (ii) displaying the camera bar's view on a display; (iii) the computer system: generating a pattern of non-random marker positions, and displaying them with the camera bar's view; (iv) detecting that a marker has been placed in a position of a non-random marker position; (v) recording a marker and a respective position of the marker; (vi) repeating steps (iv) and (v) until a predetermined number of different markers and respective marker positions have been recorded; (vii) matching markers detected with the recorded markers and their respective marker positions, to obtain a pose of the camera bar; (viii) repeating step (vii), to track the pose of the camera bar.

Abstract: A method for providing panning for a camera monitor system (CMS) includes displaying image feeds from a plurality of CMS cameras on a plurality of CMS displays. The image feeds depict an environment surrounding a commercial vehicle. The method also includes providing a human-machine interface (HMI) on a touchscreen display, receiving at least one panning command through the HMI, and, based on the at least one panning command, panning the image feed on one of the plurality of CMS displays from a default view to a panned view. A camera monitor system (CMS) is also disclosed.

Abstract: A parking facility employs an entrance camera and an exit camera to identify vehicles entering and exiting the parking facility in order to manage parking events. A secondary camera is located remote from the parking facility and nearby a hotel, restaurant or other venue where there exists a valet or check-in area. Each of the entrance, exit and secondary cameras are networked with a central computing system. Vehicles passing by the secondary camera are identified and a temporary parking permit for the parking facility can be quickly established based upon the data collected by the secondary camera so that hourly charges are not incurred when the vehicle parks in the parking facility.

Abstract: System and method for collection, control and display of cameras and images. Software may direct the cameras to targeted images automatically or manually. Artificial Intelligence (AI) may be used to identify targets and automatically direct one or more cameras to focus on a target. Images taken by one or more cameras may then be stored and displayed for later viewing or live streamed. AI may be used to analyze images (Hits) to (i) determine if such Hits are objects of interest; (ii) determine if the Hits are related to an existing Incident; (iii) if determining a Hit is related to an Incident, create a new Filter and associating the new Filter with the Incident; and (iv) if determining the Hit is unrelated to an Incident, create a new Filter, associate the Hit with the Filter and use such information to train the artificial intelligence.

Abstract: A video data processing method includes generating, based on a spliced texture map and depth map, an encoded video bitstream of a free-view video based on multiple images collected by an ith camera in a camera array. The method further includes obtaining boundary metadata information indicating a scene boundary of the free-view video, and adding the boundary metadata information to a file encapsulation data box, to obtain an extended data box. The method further includes performing encapsulation on the encoded video bitstream and the extended data box to obtain a video media file of the free-view video including the boundary metadata information. The method further includes delivering the video media file to a video client, enabling the video client to determine that the ith camera is a boundary camera, and to render video content in the scene boundary indicated by the boundary metadata information.

Abstract: In the present application, a system for composing, editing, and sharing multimedia content is disclosed. A specification of an edit to a portion of a video is received on a client. A preview of the video with the edit applied to the portion of the video is provided on the client prior to encoding the video. Input data is provided to a machine learning model, wherein the machine learning model is used to make a determination of applying the edit to the portion of the video on the client or a server of a cloud service. In response to the determination of applying the edit to the portion of the video on the server of the cloud service, at least the specification of the edit to the portion of the video and the portion of the video without the edit applied are provided to the server of the cloud service.

Abstract: A first apparatus comprises a first receiver (301) receiving images of a scene captured and a second receiver (303) receives 3D spatial data for the scene A view synthesis neural network (307) generates view shifted images for the scene for different view poses from the images and the spatial data. A neural network trainer (309) trains the view synthesis neural network (307) based on images of the scene for different view poses. A generator (305) generates an audiovisual data stream comprising: image data for the images, scene data representing the three dimensional spatial data, and coefficient data describing coefficients of the view synthesis neural network (307) after training. A second apparatus receives the audio visual data stream and sets a local neural network (403) based on the coefficient data. The local neural network (403) is then used to generate images of the scene for different view poses.

Abstract: A method for reading and writing frame images having variable frame rates and a system therefor are provided. The system provides a video processing system that includes a write end and a read end for accessing a single buffer block. In the method performed in the video processing system, when the read end starts to read an image frame, the read end is controlled to not read data from the buffer block until the data written by the write end to the buffer block reaches a progress threshold; and when it is determined that a portion of the data in the buffer block not yet read by the read end will be overwritten by the data written by the write end, the writing process is terminated until the read end finishes reading the data in the portion in the buffer block.

Abstract: A media player providing real time rewind playback of a played media file having segments of frames. A last segment N of the played media file is cached and rendered on a device, such as a mobile device, then a previous segment N?1 is cached and rendered, and the process continues until there are no more segments of the played media file to cache and render. Only a segment of the played media file is cached at a time, rather than the whole media file, such that the played media file can be replayed on the fly.

Abstract: An image processing device is disclosed that includes an image acquisition unit configured to acquire a captured image captured by an imaging device fixed to a moving object; a behavior data acquisition unit configured to acquire behavior data indicating behavior of the moving object when moving; a determination unit configured to determine whether or not the captured image includes scenery outside the moving object; and an image processing unit configured to, when the captured image is determined to include the scenery outside the moving object, perform shake correction processing on the captured image based on the behavior data acquired by the behavior data acquisition unit so as to cancel the shaking of the captured image due to the behavior.

Abstract: The present disclosure causes live content to be drawn on the basis of specification data on a virtual object, operation data on the virtual object, and the like, and causes the drawn content to be provided for viewing reproduction on a terminal for viewing.

Abstract: A video may include visual content having a progress length. A user may interact with a mobile device to set framings of the visual content at moments within the progress length. The framings of the visual content may be provided to a video editing application. The video editing application may utilize the framings set via the mobile device to provide preliminary framings of the visual content at the moments within the progress length.

Abstract: The work machine control method includes displaying a display screen including information related to a work machine, on a display device, and switching, when a switching operation unit is operated, the display screen to be displayed on the display device, among a plurality of screens according to a switching pattern. Here, the control method further comprises changing the switching pattern according to a movement state of the work machine.

Abstract: A poultry stun detection system is designed for using with a poultry slaughter platform, wherein the poultry stunning detection system includes at least: a first camera device, a second camera device, a network transmission device, and a computer processing device. The poultry stun detection system utilizes visual devices to immediately determine whether the stunned poultry still has perception in real time, and employs the YOLOv4 deep learning model to enhance the recognition accuracy of whether the poultry has lost its consciousness after stunning.

Abstract: A file decapsulation method for a free viewpoint video includes receiving a media file of free viewpoint video data. The media file includes a video track, the free viewpoint video data includes video data of N viewpoints, and the video track includes codec independence indication information and video bitstreams of M viewpoints. The codec independence indication information indicates whether video data of one of the M viewpoints in the video track depends on video data of other viewpoints during codec. The method further includes decapsulating the media file according to the codec independence indication information, to obtain a video bitstream corresponding to at least one of the M viewpoints, and decoding the video bitstream corresponding to the at least one of the M viewpoints, to obtain reconstructed video data of the at least one of the M viewpoints.