Abstract: Features are disclosed for a security camera including automated identification of disabled privacy indicators based on feedback included in monitored data (e.g., images). The features include generating a random number and adjusting light emitted by the privacy indicator to signal the random number. The camera may detect subtle changes (e.g., brightness, color, hue, intensity) in the image, a sequence of images, or portions thereof that can be decoded and compared to the random number. If there is a discrepancy between the random number and decoded value, an indication of tampering with the privacy indicator may be provided.

Abstract: Techniques are generally described for bit rate stabilization. In various examples, an encoder may identify a first frame of video data to be encoded using a first quantization value. In some examples, the encoder may determine an estimated compressed frame size of the first frame encoded using the first quantization value. In various examples, the encoder may determine that the estimated compressed frame size is less than threshold size. The encoder may further determine a second quantization value that may be less than the first quantization value. The encoder may generate a compressed first frame of video data using the second quantization value and the first frame of video data. The compressed first frame may be sent to a recipient computing device.

Abstract: Devices, systems and methods are disclosed that detect and interpret the meaning of visual and/or audio indicators of devices, and provide information, solutions to potential malfunctions, and/or automatic ordering of consumable products to users. For example, sequences of visual indicators (such as LEDs) and/or audio indicators (such as speakers that output beeps) generated by home electronics and other devices may be monitored using an image capture component (such as a camera) or audio capture component (such as a microphone), and the sequences may be interpreted to provide users with information of what is wrong with a device, how to fix a device, purchase a new device, offer repair services, etc. The systems and methods may also provide users with notifications relating to such indicators when the user is away from home.

Abstract: Devices, systems and methods are disclosed for improving color resolution in YUV subsampled image signals. Chrominance data may be subsampled using two different techniques and the two different outputs may be interlaced to improve a resulting color resolution as perceived by a viewer. Luminance data may be adjusted in paired frames to improve the perceived color resolution without affecting the perceived luminance values. High edge areas, including high frequency variations in luminance values, may be adaptively desaturated to improve the perceived color resolution of the resulting output.

Abstract: Devices and techniques are described for validation of depth data. A first pattern and second pattern may be projected. A first image of the first pattern and a second image of the second pattern may be captured. A first code word may be determined for a first pixel address based on a first value of the first pixel address in the first pattern and a second value of the first pixel address in the second pattern. A third pattern may be projected. A second code word may be determined for the first pixel address based on a third value of the first pixel address in the third pattern and the second value of the first pixel address in the second pattern. A confidence value of the first pixel address may be assigned based on the first code word and the second code word corresponding to the same projector column.

Abstract: Devices, systems and methods are disclosed that detect and mask visual artifacts in image and video data. Such artifacts may result from stitching segments of images from two different cameras due to the parallax effect. In this respect, stitching errors may be determined and compared to a threshold to decide whether an area results in an undesirable artifact that should be masked. Such artifacts may then be masked with various graphical elements such as artistic type objects/images to provide a pleasant looking image to a viewer.

Abstract: Techniques are generally described for compression encoding of images. In various examples, the techniques may comprise receiving image data. In some examples, the techniques may further comprise generating a first compressed representation of luminance values of the image data by encoding luminance values of the image data with reference to a luminance palette. In some examples, the techniques may further include generating a second compressed representation of chrominance values of the image data by applying a frequency domain based image transform to the chrominance values of the image data. In various further examples, the techniques may further comprise combining the first compressed representation and the second compressed representation to generate compressed image data.

Abstract: Devices, systems and methods are disclosed for improving color resolution in YUV subsampled signals. Chrominance data may be subsampled using two different techniques and the two different outputs may be interlaced to improve a resulting color resolution as perceived by a viewer. Luminance data may be adjusted in paired frames to improve the perceived color resolution without affecting the perceived luminance values. High edge areas, including high frequency variations in luminance values, may be adaptively desaturated to improve the perceived color resolution of the resulting output.

Abstract: The techniques described herein segment objects represented in images using one or more depth maps, regardless of whether an initial depth map is of a threshold quality. To do so, the techniques determine whether the initial depth map is of the threshold quality. If so, the techniques use this depth map for performing segmentation on an image. If not, then the techniques utilize one or more fallback approaches for generating an additional depth map, from which the segmentation may be performed.

Abstract: The techniques described herein segment objects represented in images using one or more depth maps, regardless of whether an initial depth map is of a threshold quality. To do so, the techniques contemporaneously generate, by a depth sensor, a depth map of an environment and capture, by an RGB camera, a color image of the environment. The techniques then determine whether the initial depth map is of the threshold quality. If so, the techniques use this depth map for performing segmentation on an image. If not, then the techniques generate a new depth in part by comparing an image captured by the depth sensor to the color image and performing block-matching between these images. That is, the techniques use the depth sensor and the RGB camera as stereo pair and fill in holes of the initial depth map based on the block-matching.

Abstract: A computer system determines a metric for an input object, which could be an image of a person with the metric being measure of the person's body size, age, etc. A paired neural network system is trained on a training set of objects having pairs of objects each assigned a relative metric. A relative metric for a pair indicate which of the pair has the higher metric. A representative set of objects includes a known assigned metric value for each object. The trained paired neural network system pairwise compares an input object with objects from the representative set to determine a relative metric for each such pair, to arrive at a collection of relative metrics of the input object relative to various objects in the representative set. A metric value can be estimated for the input object based on the collection of relative metrics and those known metric values.

Abstract: A video capture device may include multiple cameras that simultaneously capture video data. The video capture device and/or one or more remote computing resources may perform video stabilization to the video data to reduce unwanted camera shake. The video capture device may include one or more motion sensors that track the motion of the video capture device. Using the motion data, video stabilization algorithms can be executed on a first video stream to calculate stabilizing adjustments for the video stream. The stabilization adjustments can then be used to stabilize a second video stream due to the fixed positioning of a first camera that captured the first video stream and a second camera that captured the second video stream. The video capture device and/or the remote computing resources may stitch together the first and second video streams to generate a digitally stabilized panoramic video.

Abstract: Devices, systems and methods are disclosed that detect and interpret the meaning of visual indicators (such as LEDs) of devices, and provide information and solutions to potential malfunctions to users. For example, image data of the visual indicators can be captured by an image capture device (such as a camera). The image data may then be analyzed to identify an object captured in the image data. Luminescence of pixel values in the image data corresponding to a visual indicator of the object may be used to determine a sequence of the visual indicator. Then, information corresponding to the meaning of the sequence may be sent to a device of a user or owner of the object.

Abstract: Various examples are directed to systems and methods for formatting stitched panoramic frames for transmission to a remote image processing system, the method comprising. A control circuit of a panoramic camera may receive first and second frames captured by first and second image sensors having partially overlapping fields-of-view. The control circuit may identify a first overlap region of the first frame and a second overlap region of the second frame. The first and second overlap region may comprise pixel values from portions of the fields-of-view of the image sensors that overlap. The control circuit may generate a panoramic frame from based at least in part on the first and second frames. The control circuit may generate a transmission frame comprising the panoramic frame, the first overlap region and the second overlap region.

Abstract: Devices, systems and methods are disclosed for improving a field of view of video data by combining video data from multiple image capture devices. For example, video data from a first image capture device and a second image capture device located above or below the first image capture device may be combined to increase a field of view. The boundary between the first video data and the second video data may be determined based on objects visible in the video data. In addition, if the first image capture device and the second image capture device include two or more cameras, the vertical overlap between the first image capture device and the second image capture device may be used to horizontally stitch video data from the two or more cameras.

Abstract: An image processor merges images from a plurality of cameras including applying spatially-varying gains over images from individual cameras to obtain a consistent effective where the images are stitched together despite the cameras possibly having different camera exposures. The image processor can adjust the camera exposures to improve stitching results. In addition to gain modification, the image processor can also modify the images for other artifacts, such as veiling glare and make those less visible at the seams of images. The image processor also can take into account constraints on gain changes and optimize camera parameters to minimize stitching artifacts. Using a cost function, the image processor can optimize for constraints on camera exposure that would otherwise result in camera exposure differences that would themselves cause visible artifacts.

Abstract: Devices, systems and methods are disclosed for improving encoding techniques for mobile devices by adaptively controlling a resolution or frame rate of content to reduce power consumption while maintaining image quality. For example, a local device may determine when the content may be downscaled without degrading a final image quality and may downscale the content prior to encoding and transmitting the encoded content to a remote device. The remote device may decode and upscale the content to the original resolution prior to displaying the content on a display. As downscaling the content is not power intensive, the local device may reduce a power consumption associated with encoding and transmitting the content to the remote device while maintaining the final image quality of the content.

Abstract: Various examples are directed to an image processing system comprising a mounting assembly, a first image sensor, a second image sensor, and a third image sensor. A processor may be programmed to receive, from the first image sensor, a first frame, receive, from the third image sensor, a third frame, and identify a first optical flow from the first frame to the third frame. Based at least in part on the first optical flow, the processor may generate a translation kernel describing a first translation from a first position of the first frame to a destination position, and apply the translation kernel to generate a translated first frame.

Abstract: An electronic device can have two or more pairs of cameras capable of performing three-dimensional imaging. In order to provide accurate disparity information, these cameras should be sufficiently calibrated. Automatic calibration can be performed by periodically capturing images with a pair of front-facing cameras and locating matching facial or other feature points in corresponding images captured by those cameras. Correspondences can be detected between feature points and the corresponding feature points can be normalized and outlier feature points can be rejected. A transformation matrix can be determined using at least a portion of remaining feature points and can be used to determine rotation and translation parameters to correct for misalignment between the cameras. The calibration parameters can be refined or otherwise adjusted, and can be used or stored for use in correcting images subsequently captured by those cameras.

Abstract: Various examples described herein are directed to systems and methods for stabilizing a panoramic video. The panoramic video may comprise a first frame captured by a panoramic camera. The panoramic camera may be positioned in a three-dimensional camera space described by an x-axis, a y-axis, and a z-axis that are mutually orthogonal. The first frame may comprise a plurality of pixel values arranged in a pixel plane and may be continuous in a first dimension. An image processor may receive from a motion sensor data describing a rotation of the panoramic camera at a time when the panoramic camera captured the first frame. The image processor may determine a planar component of an unintended rotation of the panoramic camera about the z-axis and convert the planar component of the unintended rotation to a first frame shift in the first dimension.