Abstract: A system and method are provided for training a machine learning system. In an embodiment, the system generates a three-dimensional model of an environment using a video sequence that includes individual frames taken from a variety of perspectives and environmental conditions. An object in the environment is identified and labeled, in some examples, by an operator, and a three-dimensional model of the object is created. Training data for the machine learning system is created by applying the label to the individual video frames of the video sequence, or by applying a rendering of the three-dimensional model to additional images or video sequences.

Abstract: A system and method are provided for training a machine learning system. In an embodiment, the system generates a three-dimensional model of an environment using a video sequence that includes individual frames taken from a variety of perspectives and environmental conditions. An object in the environment is identified and labeled, in some examples, by an operator, and a three-dimensional model of the object is created. Training data for the machine learning system is created by applying the label to the individual video frames of the video sequence, or by applying a rendering of the three-dimensional model to additional images or video sequences.

Abstract: Automatic organ localization is described. In an example, an organ in a medical image is localized using one or more trained regression trees. Each image element of the medical image is applied to the trained regression trees to compute probability distributions that relate to a distance from each image element to the organ. At least a subset of the probability distributions are selected and aggregated to calculate a localization estimate for the organ. In another example, the regression trees are trained using training images having a predefined organ location. At each node of the tree, test parameters are generated that determine which subsequent node each training image element is passed to. This is repeated until each image element reaches a leaf node of the tree. A probability distribution is generated and stored at each leaf node, based on the distance from the leaf node's image elements to the organ.

Abstract: Methods for mapping color data having at least one color associated therewith to an output device based on an input-device profile and an output-device profile, each profile having a tone curve and a color matrix, are provided. In one embodiment, the method includes receiving color data from an input device and determining whether the color data is in a linear space. If it is determined that the color data is not in a linear space, the method further includes applying the tone curve of the input device profile to the color data to convert it into a linear space. The method further includes converting the color(s) associated with the color data from the input linear space to an output linear space by applying the color matrix of the input device profile and the inverse color matrix of the output device profile to create color-converted image data.

Abstract: Methods and a processing device are provided for restoring pixels damaged by artifacts caused by dust, or other particles, entering a digital image capturing device. A user interface may be provided for a user to indicate an approximate location of an artifact appearing in a digital image. Dust attenuation may be estimated and an inverse transformation, based on the estimated dust attenuation, may be applied to damaged pixels in order to recover an estimate of the underlying digital image. One or many candidate source patch may be selected based on having smallest pixel distances, with respect to a target patch area. The damaged pixels included in the target patch area may be considered when calculating the pixel distance with respect to candidate source patches. RGB values of corresponding pixels of source patches may be used to restore the damaged pixels included in the target patch area.

Abstract: Automatic organ localization is described. In an example, an organ in a medical image is localized using one or more trained regression trees. Each image element of the medical image is applied to the trained regression trees to compute probability distributions that relate to a distance from each image element to the organ. At least a subset of the probability distributions are selected and aggregated to calculate a localization estimate for the organ. In another example, the regression trees are trained using training images having a predefined organ location. At each node of the tree, test parameters are generated that determine which subsequent node each training image element is passed to. This is repeated until each image element reaches a leaf node of the tree. A probability distribution is generated and stored at each leaf node, based on the distance from the leaf node's image elements to the organ.

Abstract: Methods for mapping color data having at least one color associated therewith to an output device based on an input-device profile and an output-device profile, each profile having a tone curve and a color matrix, are provided. In one embodiment, the method includes receiving color data from an input device and determining whether the color data is in a linear space. If it is determined that the color data is not in a linear space, the method further includes applying the tone curve of the input device profile to the color data to convert it into a linear space. The method further includes converting the color(s) associated with the color data from the input linear space to an output linear space by applying the color matrix of the input device profile and the inverse color matrix of the output device profile to create color-converted image data.

Abstract: Methods and a processing device are provided for restoring pixels damaged by artifacts caused by dust, or other particles, entering a digital image capturing device. A user interface may be provided for a user to indicate an approximate location of an artifact appearing in a digital image. Dust attenuation may be estimated and an inverse transformation, based on the estimated dust attenuation, may be applied to damaged pixels in order to recover an estimate of the underlying digital image. One or many candidate source patch may be selected based on having smallest pixel distances, with respect to a target patch area. The damaged pixels included in the target patch area may be considered when calculating the pixel distance with respect to candidate source patches. RGB values of corresponding pixels of source patches may be used to restore the damaged pixels included in the target patch area.

Abstract: Methods and systems for processing digital image data utilizing vertically-oriented Effect graphs are provided. When processing digital image data utilizing Effect graphs, it is often necessary for certain Effects on the graph to render their outputs multiple times during a single rendering pass. To alleviate the exponential processing that such a scenario can cause, methods and systems for caching at least a portion of the digital image data being processed in image buffers associated with the output of one or more Effects in the Effect graph during processing are provided. Additionally provided are methods and systems for caching digital image data in image buffers associated with the output of one or more Effects across multiple processing passes of an Effect graph.

Abstract: Systems for rendering Effect graphs for non-destructively processing digital image data which integrate Central Processing Unit (CPU) processing and Graphics Processing Unit (GPU) processing are provided. Additionally provided are systems for processing digital image data utilizing Effect graphs. The systems of the present invention integrate CPU processing and GPU processing to facilitate accelerated rendering of Effect graphs and, consequently, accelerated processing of digital images. Methods for processing digital image data utilizing the systems herein described are also provided.

Abstract: Technologies, methods, and systems for generating and maintaining profile information with processed image data, and for enabling scene-referred high-fidelity adjustments to non-raw, processed image data.

Abstract: Methods and systems for processing, e.g., non-destructively processing, digital image data utilizing vertically-oriented Effect graphs are provided. In non-destructive processing where and when data is transformed is fairly important, both in terms of quality and performance. The further down the vertically-oriented Effect graph a transformation occurs, the better. As such, methods for pushing transformations down an Effect graph to the lowest point possible and applying them at that point rather than the location at which they may have been placed are provided. Systems for implementing the methods herein disclosed are also provided.

Abstract: Methods and systems for processing, e.g., non-destructively processing, digital image data utilizing one or more Effect Layers are provided. Effect Layers combine Effects in useful ways that simplify the process of creating Effect graphs. Each Effect Layer contains a logical Effect sub-graph that includes a plurality of logical Effects, e.g., a main Effect, a blend Effect, and a mask Effect. In one embodiment, a method in accordance with an embodiment of the present invention includes receiving input regarding processing of digital image data, determining the impact that input will have on each logical Effect upon processing of digital image data, and creating a physical Effect sub-graph in accordance with the determined impact. The physical Effect sub-graph may closely resemble or appear nothing like the logical Effect sub-graph. Images may subsequently be rendered utilizing the physical Effect sub-graph.