Abstract: A media item to be provided to users of a platform is identified. The media item is associated with a media class of one or more media classes. An indication of the media item is provided as input to a machine learning model trained based on historical encoding data to predict, for a given media item, a set of encoder parameter settings that satisfy a performance criterion in view of a respective media class of the given media item. The historical encoding data includes a prior set of encoder parameter settings that satisfied the performance criterion with respect to a prior media item associated with the respective class. Encoder parameter settings that satisfy the performance criterion in view of the media class is determined based on an output of the model. The media item is caused to be encoded using the determined encoder parameter settings.

Abstract: Methods and systems for encoder parameter setting optimization. A media item to be provided to one or more users of a platform is identified. The media item is associated with a media class. An indication of the identified media item is provided as input to a first machine learning model. The first machine learning model is trained to predict, for a given media item, a set of encoder parameter settings that satisfy a performance criterion in view of a respective media class associated with the given media item. One or more outputs of the first machine learning model are obtained. The one or more obtained outputs include encoder data identifying one or more sets of encoder parameter settings and, for each of the sets of encoder parameter settings, an indication of a level of confidence that a respective set of encoder parameter settings satisfies the performance criterion in view of the media class associated with the identified media item.

Abstract: Methods and systems for encoder parameter setting optimization. A media item to be provided to one or more users of a platform is identified. The media item is associated with a media class. An indication of the identified media item is provided as input to a first machine learning model. The first machine learning model is trained to predict, for a given media item, a set of encoder parameter settings that satisfy a performance criterion in view of a respective media class associated with the given media item. One or more outputs of the first machine learning model are obtained. The one or more obtained outputs include encoder data identifying one or more sets of encoder parameter settings and, for each of the sets of encoder parameter settings, an indication of a level of confidence that a respective set of encoder parameter settings satisfies the performance criterion in view of the media class associated with the identified media item.

Abstract: A first aspect is a method for coding a group of pictures (GOP) that includes frames of a video. The method includes encoding, at least some of the frames of the GOP, using a first encoding pass to obtain encoding statistics; obtaining, using the encoding statistics, respective temporal dependency likelihoods (TDLs) for the at least some of the frames of the GOP, where the respective TDLs indicate contributions that the at least some of the frames make in reducing prediction errors of the GOP; obtaining a reference frame based on the respective TDLs; and using the reference frame in encoding at least some of the frames of the GOP in a second encoding pass.

Abstract: A first aspect is a method for coding a group of pictures (GOP) that includes frames of a video. The method includes encoding, at least some of the frames of the GOP, using a first encoding pass to obtain encoding statistics; obtaining, using the encoding statistics, respective temporal dependency likelihoods (TDLs) for the at least some of the frames of the GOP, where the respective TDLs indicate contributions that the at least some of the frames make in reducing prediction errors of the GOP; obtaining a reference frame based on the respective TDLs; and using the reference frame in encoding at least some of the frames of the GOP in a second encoding pass.

Abstract: Embodiments of the present invention generate estimates of device motion from two data sources on a computing device—a motion sensor and a camera. The device may compare the estimates to each other to determine if they agree. If they agree, the device may confirm that device motion estimates based on the motion sensor are accurate and may output those estimates to an application within the device. If the device motion estimates disagree, the device may alter the motion estimates obtained from the motion sensor before outputting them to the application.

Abstract: Implementations disclose bitrate optimization for multi-representation encoding using playback statistics.

Abstract: A method includes determining whether a rate distortion cost of a compressed video is above a cost threshold, the compressed video being encoded using a first constant rate factor (CRF). Upon determining the quality of a compressed video is above a cost threshold calculating a second CRF based on the first CRF, and encoding a video associated with the compressed video using the second CRF. Upon determining the quality of a compressed video is below a cost threshold encoding the video using the first CRF and a target bitrate.

Abstract: Systems and methods for processing a video sequence are disclosed. In accordance with some implementations, The method includes determining a first set of shot-change locations associated with a cut transition in a video sequence based on variance data defined in an encoding log associated with at least one previously encoded video frame. The method further includes determining a second set of shot-change locations associated with a fading transition in the video sequence based on the variance data defined in the encoding log associated with the at least one previously encoded video frame. The method also includes processing the video sequence based on at least a portion of the first set of shot-change locations and the second set of shot-change locations.

Abstract: Implementations disclose bitrate optimization for multi-representation encoding using playback statistics.

Abstract: Embodiments of the present invention generate estimates of device motion from two data sources on a computing device—a motion sensor and a camera. The device may compare the estimates to each other to determine if they agree. If they agree, the device may confirm that device motion estimates based on the motion sensor are accurate and may output those estimates to an application within the device. If the device motion estimates disagree, the device may alter the motion estimates obtained from the motion sensor before outputting them to the application.

Abstract: Systems and methods for processing a video sequence are disclosed. In accordance with some implementations, The method includes determining a first set of shot-change locations associated with a cut transition in a video sequence based on variance data defined in an encoding log associated with at least one previously encoded video frame. The method further includes determining a second set of shot-change locations associated with a fading transition in the video sequence based on the variance data defined in the encoding log associated with the at least one previously encoded video frame. The method also includes processing the video sequence based on at least a portion of the first set of shot-change locations and the second set of shot-change locations.

Abstract: An instruction input apparatus is provided. The instruction input apparatus comprises a keyboard, on which a plurality of buttons are disposed; a camera module, which comprises a supporting bar and a camera, wherein a first end of the supporting bar is connected to the keyboard, the camera is disposed on a second end of the supporting bar so as to protrude the camera from the keyboard, and the camera comprises at least two lens modules for capturing a plurality of image frames; and an image processing unit, which is disposed in the keyboard and is signal-connected to the camera so that the image processing unit receives and processes the image frames for merging the image frames into a panorama image. Therefore, the instruction input apparatus would improve the functionality of the instruction input apparatus and provide function of all viewing angles photographing so that the instruction input apparatus has an advantage of taking all desired images together.

Abstract: A system to facilitate analyzing and/or processing video content using encoding logs is provided. The system includes a first analysis component, a second analysis component and a transcoding component. The first analysis component determines a first set of shot-change locations associated with a cut transition in a video sequence based on an encoding log associated with at least one previously encoded video frame. The second analysis component determines a second set of shot-change locations associated with a fading transition in the video sequence based on the encoding log associated with the at least one previously encoded video frame. The transcoding component processes the video sequence based on at least a portion of the first set of shot-change locations and the second set of shot-change locations.

Abstract: A keyboard device includes a keyboard body including a plurality of keys. A projector is disposed in the keyboard body for receiving and transforming an image signal into a projection frame, and projecting the projection frame on a projection plane. A vibration sensor is disposed in the keyboard body, and generates a vibration signal in response to a vibration of the keyboard body. A microprocessor is in communication with the vibration sensor and the projector, dynamically adjusts the image signal according to the vibration signal, and outputs the adjusted image signal to the projector to generate the projection frame corresponding to the adjusted image signal. The shaking problem of the projection frame can thus be ameliorated.

Abstract: Embodiments of the present invention generate estimates of device motion from two data sources on a computing device—a motion sensor and a camera. The device may compare the estimates to each other to determine if they agree. If they agree, the device may confirm that device motion estimates based on the motion sensor are accurate and may output those estimates to an application within the device. If the device motion estimates disagree, the device may alter the motion estimates obtained from the motion sensor before outputting them to the application.

Abstract: Video coders may perform perspective transformation of reference frames during coding in a manner that conserves processing resources. When a new input frame is available for coding, a camera position for the input frame may be estimated. A video coder may search for reference pictures having similar camera positions as the position of the input frame and, for each reference picture identified, the video coder may perform a prediction search to identify a reference picture that is the best prediction match for the input frame. Once the video coder identifies a reference picture to serve as a prediction source for the input frame, the video coder may derive a transform to match the reference frame data to the input frame data and may transform the reference picture accordingly. The video coder may code the input frame using the transformed reference picture as a prediction reference and may transmit coded frame data and the camera position of the input frame to a decoder.

Abstract: Embodiments of the present invention generate estimates of device motion from two data sources on a computing device—a motion sensor and a camera. The device may compare the estimates to each other to determine if they agree. If they agree, the device may confirm that device motion estimates based on the motion sensor are accurate and may output those estimates to an application within the device. If the device motion estimates disagree, the device may alter the motion estimates obtained from the motion sensor before outputting them to the application.

Abstract: In a video coding/decoding system, reference picture caches in a video coder and decoder may be partitioned dynamically based on camera and background motion can lead to improved coding efficiency and coding quality. When a camera is fixed and therefore exhibits low motion, a system may allocate larger portions of the reference picture cache to storage of long term reference frames. In this case, foreground elements of an image (for example, a person) may move in front of a relatively fixed background. Increasing the number of long term reference frames can increase the chances that, no matter where the foreground elements are within a frame currently being coded, the reference picture cache will contain at least one frame that provides an adequate prediction match to background elements within the new frame. Thus the background elements uncovered in the current frame can be coded at high quality with a low number of bits.

Abstract: Embodiments of the present invention generate estimates of device motion from two data sources on a computing device—a motion sensor and a camera. The device may compare the estimates to each other to determine if they agree. If they agree, the device may confirm that device motion estimates based on the motion sensor are accurate and may output those estimates to an application within the device. If the device motion estimates disagree, the device may alter the motion estimates obtained from the motion sensor before outputting them to the application.