Qualcomm Incorporated

Thomson Licensing

Gokce Dane

Khaled Helmi El-Maleh

Vasudev Bhaskaran

Chia-Yuan Teng

Cristina Gomila

Yen-Chi Lee

Min Dai

Upal Mahbub

Giovanni Motta

Liang Liang

Ashar Ali

Bob R Hung

Wuttipong Kumwilaisak

Samuel Sunarjo

Yan Ye

Xiaoan Lu

Gerhard Reitmayr

Dashan Gao

Ming-Chang Tsai

Sujith Srinivasan

Xin Zhong

Raghavendra Nagaraj

Dan MingLun Chuang

Vladan Andrijanic

Matteo Toti Mannino

Yiu-Wing Leung

Nien-Chung Feng

Karl Ni

Methods, systems, and apparatuses are provided to automatically detect objects within images. For example, an image capture device may capture an image, and may apply a trained neural network to the image to generate an object value and a class value for each of a plurality of portions of the image. Further, the image capture device may determine, for each of the plurality of image portions, a confidence value based on the object value and the class value corresponding to each image portion. The image capture device may also detect an object within at least one image portion based on the confidence values. Further, the image capture device may output a bounding box corresponding to the at least one image portion. The bounding box defines an area of the image that includes one or more objects.

Apparatus and methods for object detection using machine learning processes

Systems and techniques are provided for registering three-dimensional (3D) images to deformable models. An example method can include determining, based on an image of a target and associated depth information, a 3D mesh of the target; determining different sets of rotation and translation parameters based on modifications to rotation and translation parameters of the 3D mesh; generating, based on the different sets of rotation and translation parameters, different 3D meshes having different orientations, different poses, and/or different alignments relative to the target; determining different sets of model parameters associated with the different 3D meshes, based on modifications to the different sets of rotation and translation parameters; generating, based on the different sets of model parameters, different additional 3D meshes having different orientations, different poses, and/or different alignments relative to the target; and selecting a final 3D mesh of the target from the different additional 3D meshes.

Three-dimensional scan registration with deformable models

Systems and techniques are provided for modeling three-dimensional (3D) meshes using images. An example method can include receiving, via a neural network system, an image of a target and metadata associated with the image and/or a device that captured the image; determining, based on the image and metadata, first 3D mesh parameters of a first 3D mesh of the target, the first 3D mesh parameters and first 3D mesh corresponding to a first reference frame associated with the image and/or the device; and determining, based on the first 3D mesh parameters, second 3D mesh parameters for a second 3D mesh of the target, the second 3D mesh parameters and second 3D mesh corresponding to a second reference frame, the second reference frame including a 3D coordinate system of a real-world scene where the target is located.

Modeling objects from monocular camera outputs

A method performed by an electronic device is described. The method includes generating a depth map of a scene external to a vehicle. The method also includes performing first processing in a first direction of a depth map to determine a first non-obstacle estimation of the scene. The method also includes performing second processing in a second direction of the depth map to determine a second non-obstacle estimation of the scene. The method further includes combining the first non-obstacle estimation and the second non-obstacle estimation to determine a non-obstacle map of the scene. The combining includes combining comprises selectively using a first reliability map of the first processing and/or a second reliability map of the second processing The method additionally includes navigating the vehicle using the non-obstacle map.

Systems and methods for non-obstacle area detection

A method performed by an electronic device is described. The method includes obtaining a motion vector map based on at least two images. The motion vector map has fewer motion vectors than a number of pixels in each of the at least two images. The method also includes obtaining a feature point from one of the at least two images. The method further includes performing a matching operation between a template associated with the feature point and at least one search space based on the motion vector map. The method additionally includes determining a motion vector corresponding to the feature point based on the matching operation.

Systems and methods for determining feature point motion

A method performed by an electronic device is described. The method includes performing vertical processing of a depth map to determine a vertical non-obstacle estimation. The method also includes performing horizontal processing of the depth map to determine a horizontal non-obstacle estimation. The method further includes combining the vertical non-obstacle estimation and the horizontal non-obstacle estimation. The method additionally includes generating a non-obstacle map based on the combination of the vertical and horizontal non-obstacle estimations.

This disclosure describes methods and apparatus for decoding data. In one aspect, the method comprises decoding encoded video data to obtain decoded video frame data, the encoded video data comprising encoded video frame data encoded at a first frame rate and embedded data. The method further comprises determining a camera parameter from the embedded data and up-converting the decoded video frame data to a second frame rate based on the camera parameter. The determined camera parameter may be, for example, a parameter associated with one or more of a zoom factor, an auto focus status, lens position information, frame luma information, an auto exposure (AE) convergence status, an automatic white balance (AWB) convergence status, global motion information, and frame blurriness information, and the like. An encoding device may embed the camera parameter(s) in an encoded video bit stream for a decoder to utilize during frame rate up-conversion.

Camera parameter-assisted video frame rate up conversion

A system and method for correcting artifacts present in video frames is disclosed. The system includes a decision module () and an artifact correction unit (). The system receives artifact metrics () and artifact maps () corresponding to artifacts present in video frames and prioritizes the video frames to be corrected by considering the type and severity of the artifacts and the content of the video frames (). Subsequent to prioritizing the video frames, the system corrects artifacts of selected frames by adjusting compressed domain parameters (), such as quantization parameters and mode decisions. An encoder () compresses the video frames according to the adjusted compressed domain parameter to automatically provide a video stream with a lower incidence and severity of artifacts.

Method and apparatus for automatic visual artifact analysis and artifact reduction

A method performed by an electronic device is described. The method includes generating a plurality of bounding regions based on an image. The method also includes determining a subset of the plurality of bounding regions based on at least one criterion and a selected area in the image. The method further includes processing the image based on the subset of the plurality of bounding regions.

Systems and methods for reducing a plurality of bounding regions

In one example, a method includes identifying a pixel in an image frame that is a candidate for causing crosstalk between the image frame and a corresponding image frame in a multiview image system. The method further includes, for a pixel identified as a candidate for causing crosstalk, applying crosstalk correction to the pixel. The method further includes applying a location-based adjustment to the pixel, wherein the location-based adjustment is based at least in part on which of two or more portions of the image frame the pixel is in.

Crosstalk reduction with location-based adjustment in multiview video processing

The disclosure is directed to techniques for encoder-assisted adaptive interpolation of video frames. According to the disclosed techniques, an encoder generates information to assist a decoder in interpolation of a skipped video frame, i.e., an S frame. The information permits the decoder to reduce visual artifacts in the interpolated frame and thereby achieve improved visual quality. The information may include interpolation equation labels that identify selected interpolation equations to be used by the decoder for individual video blocks. As an option, to conserve bandwidth, the equation labels may be transmitted for only selected video blocks that meet a criterion for encoder-assisted interpolation. Other video blocks without equation labels may be interpolated according to a default interpolation technique.

Encoder-assisted adaptive video frame interpolation

This disclosure describes selection of reference video units to be used for interpolation or extrapolation of a video unit, such as a video frame. A decoder may apply a quality-focused mode to select a reference frame based on quality criteria. The quality criteria may indicate a level of quality likely to be produced by a reference frame. If no reference frames satisfy the quality criteria, interpolation or extrapolation may be disabled. Display of an interpolated or extrapolated frame may be selectively enabled based on a quality analysis. A decoder may apply a resource-focused frame interpolation mode to enable or disable frame interpolation or extrapolation for some frames based on power and quality considerations. In one mode, frame interpolation may be disabled to conserve power when reference frames are not likely to produce satisfactory quality. In another mode, the threshold may be adjustable as a function of power saving requirements.

Reference selection for video interpolation or extrapolation

There are provided methods and apparatus for enhanced performance in a multi-pass video encoder. An apparatus includes a video encoder for encoding a video bitstream. The apparatus further includes a video quality analyzer, in signal communication with the encoder, for performing a video quality analysis of the video bitstream for a given encoding pass to detect encoder compression artifacts in the video bitstream, and providing to the encoder information relating to the encoder compression artifacts to enable a re-encoding setup of the encoder to reduce an occurrence of the encoder compression artifacts in subsequent re-encoding passes. At least one of the video quality analysis and the re-encoding setup are automated.

Methods and apparatus for enhanced performance in a multi-pass video encoder

This disclosure describes techniques for obtaining one or more video capture parameters associated with a video capture device and using the obtained video capture parameters to control a video decoding process. The obtained video capture parameters may be, for example, parameters associated with auto exposure control (AE), auto focus (AF), and auto white balance (AWB), zooming, motion in captured frames, and the like. The video encoding device may embed the video capture parameters in the encoded video bit stream such that a decoder may utilize the camera parameters during frame rate up-conversion.

Growing community of inventors

San Diego, CA, United States of America

Gokce Dane