Abstract: A face located along a stitch line in a spherical image is detected by rendering views of regions of the spherical image along the stitch line. The spherical image may be produced by combining first and second images. A first view of a projection of the spherical image is rendered. A scaling factor for rendering a second view of the projection is determined based characteristics of the first portion of the face. The second view is then rendered according to the scaling factor. The use of the scaling factor to render the second view causes a change in the depiction of the second portion of the face. For example, the scaling factor can indicate to change the resolution or expected size of the second portion of the face when rendering the second view. A face is then detected within the spherical image based on the rendered first and second views.

Abstract: Scene-based automatic white balance correction may be performed by obtaining an input image. A histogram may be computed from an image thumbnail. The scene-based automatic white balance correction may include classifying a scene. The scene-based automatic white balance correction may include learning a first filter and a second filter. The first filter and the second filter may be learned from one or several different instances of the raw image thumbnail, the augmented image thumbnail, the scene classification, or any combination thereof. The scene-based automatic white balance correction may include applying the filter to the histogram to determine white balance correction coefficients and obtain a processed image.

Abstract: Methods and systems for denoising. A method for denoising includes receiving an image from an image sensor, denoising the image, in a non-linear domain by a denoiser, by applying a noise map to the image to obtain a denoised image. Training losses for the denoiser are processed in a linear domain. The method includes storing, displaying, or transmitting an output image based on the denoised image.

Abstract: Systems and methods are used for calibrating a time difference between an image signal processor (ISP) and an inertial measurement unit (IMU) of an image capture device. An image capture device includes a lens, an image sensor, an IMU, and an ISP. The image sensor detects images as frames and the IMU captures motion data. The ISP detects one or more key points on the frames and matches the one or more key points between the frames. The ISP computes one or more calibration parameters. The one or more calibration parameters are based on the matched key points and a time difference between the ISP and the IMU. The ISP performs a calibration using the calibration parameters.

Abstract: Holistic systems and methods are used for calibrating image capture devices. An image capture device includes a lens, an image sensor, an inertial measurement unit (IMU), and an image signal processor (ISP). The image sensor detects images as frames and the IMU captures motion data. The ISP detects one or more key points on the frames and matches the one or more key points between the frames. The ISP computes one or more calibration parameters. The one or more calibration parameters are based on the matched key points and a model. The model includes an optical component, an IMU component, and a sensor component. The ISP performs a calibration using the calibration parameters.

Abstract: A method and apparatus may be used for performing a scene-based automatic white balance correction. The method may include obtaining an input image. The method may include obtaining a raw image thumbnail. The method may include obtaining an augmented image thumbnail. The method may include computing a histogram from an image thumbnail. The method may include determining a scene classification. The method may include learning a filter. The filter may be learned from one or several different instances of the raw image thumbnail, the augmented image thumbnail, the scene classification, or any combination thereof. The method may include applying the filter to the histogram to determine white balance correction coefficients and obtain a processed image.

Abstract: A face located along a stitch line in a spherical image is detected by rendering views of regions of the spherical image along the stitch line. The spherical image may be produced by combining first and second images. A first view of a projection of the spherical image is rendered. A scaling factor for rendering a second view of the projection is determined based characteristics of the first portion of the face. The second view is then rendered according to the scaling factor. The use of the scaling factor to render the second view causes a change in the depiction of the second portion of the face. For example, the scaling factor can indicate to change the resolution or expected size of the second portion of the face when rendering the second view. A face is then detected within the spherical image based on the rendered first and second views.

Abstract: A method and apparatus may be used for performing a scene-based automatic white balance correction. The method may include obtaining an input image. The method may include obtaining a raw image thumbnail. The method may include obtaining an augmented image thumbnail. The method may include computing a histogram from an image thumbnail. The method may include determining a scene classification. The method may include learning a filter. The filter may be learned from one or several different instances of the raw image thumbnail, the augmented image thumbnail, the scene classification, or any combination thereof. The method may include applying the filter to the histogram to determine white balance correction coefficients and obtain a processed image.

Abstract: Generating a calibrated camera alignment model for image capture devices having overlapping fields-of-view may include identifying a camera alignment model describing a first candidate alignment path for a defined location in a first input frame and a second candidate alignment path for the defined location in a second input frame, identifying a third candidate alignment path spatially adjacent to the first candidate alignment path, identifying a fourth candidate alignment path spatially adjacent to the second candidate alignment path, identifying a first point along the first candidate alignment path or the second candidate alignment path corresponding to a second point along the third candidate alignment path or the fourth candidate alignment path, and updating the camera alignment model based on the first point, the second point, or both.