Abstract: Aspects of the present invention relate to a computer-implemented training method for training a segmentation model to segment an image. The method includes receiving a training data set which includes first image data representing a first image comprising a foreground and a background. The first image data is captured by at least one first visible electromagnetic radiation imaging device. The method comprises augmenting the first image data by applying a first overlay image data to the first image to generate first composite image data. The first composite image data is processed using the segmentation model. The segmentation model is updated in dependence on the processing of the first composite image data to generate a first updated segmentation model. According to a further aspect of the present invention there is provided a system for training a segmentation model to segment an image. Aspects of the present invention also relate to an image processing system and method.

Abstract: Aspects of the present invention relate to a computer-implemented training method for training a segmentation model to segment an image. The method includes receiving a plurality of first training data sets. The first training data sets each include first image data representing a first image consisting of a background or of a foreground and first annotation data identifying the presence solely of the background or the foreground in the first image. The first image data is captured by at least one visible electromagnetic radiation imaging device. The method includes, for each first training data set, processing the first image data using the segmentation model to generate a first candidate segmentation; and supplying the first annotation data to an error calculating algorithm to determine a first error for the first candidate segmentation. The segmentation model is updated in dependence on the determined first error.

Abstract: Aspects of the present invention relate to a computer-implemented training method for training a segmentation model to segment an image. The method includes receiving a plurality of first training data sets for training the segmentation model. The first training data sets each include first image data representing a first image comprising a foreground and a background. The first image data is captured by at least one visible electromagnetic radiation imaging device. The method comprises, for each first training data set, generating a first target segmentation for differentiating between the foreground and the background of the first image. The first image data is processed using the segmentation model to segment the first image and generate a first candidate segmentation. The segmentation model compares the first candidate segmentation and the first target segmentation to determine a first error. The segmentation model is updated in dependence on the determined first error.

Abstract: Aspects of the present invention relate to a computer-implemented training method for training a segmentation model to segment an image. The method includes receiving a plurality of training data sets each including image data representing an image comprising a foreground and a background; and sample image data comprising one or more sample image occurring in the background of the image. The image data is captured by at least one visible electromagnetic radiation imaging device. The method includes processing each training data set using the segmentation model. The processing of each training data includes supplying the sample image data to the segmentation model; and segmenting the image data to generate a candidate segmentation in dependence on the sample image data. An error is determined for the candidate segmentation. The segmentation model is updated in dependence on the determined error.

Abstract: Aspects of the present invention relate to a computer-implemented training method for training a segmentation model to segment an image. The method includes receiving a plurality of training data sets each including image data representing an image comprising a foreground and a background; and a background composition data defining a composition of at least a portion of the background of the image represented by the image data. The image data is captured by at least one visible electromagnetic radiation imaging device. The method includes processing each training data set using the segmentation model to generate a candidate segmentation. The background composition data and the candidate segmentation are supplied to an error calculating algorithm to determine an error. The segmentation model is updated in dependence on the determined error. According to a further aspect of the present invention there is provided a system for training a segmentation model to segment an image.

Abstract: An image content replacement apparatus and method wherein a camera image receiving unit receives video images observing a scene including a subject and a mask signal generating unit generates a mask signal that defines marked areas of the video images corresponding to the subject. A content substitution unit substitutes the marked areas with alternate image content according to the mask signal to output modified camera images. An image selector unit selects the alternate image content amongst at least a first alternate image content when the subject is determined to be in a first condition within the scene and a second alternate image content when the subject is determined to be in a second condition within the scene. In examples, the first and second alternate image contents are selected based on a determined camera zoom value or a camera angle.

Abstract: An original image layer is mixed with one or more graphics layers to form a composite video signal, such as in a vision mixer. Also, the same graphics layers are mixed identically into another, known background image layer to give a second composite video signal. The currently applied added graphics layers are determined, without directly accessing the added graphics layers themselves, using only the first and second composite video signals, the original image layer, and the known background image, such as in a graphics layer detection device. A triangulation matting algorithm may be used in one example.

Abstract: An original image layer is mixed with one or more graphics layers to form a composite video signal, such as in a vision mixer. Also, the same graphics layers are mixed identically into another, known background image layer to give a second composite video signal. The currently applied added graphics layers are determined, without directly accessing the added graphics layers themselves, using only the first and second composite video signals, the original image layer, and the known background image, such as in a graphics layer detection device. A triangulation matting algorithm may be used in one example.

Abstract: An image content replacement apparatus and method wherein a camera image receiving unit (44) receives video images (21) observing a scene including a subject (10) and a mask signal generating unit (46) generates a mask signal (43) that defines marked areas of the video images (21) corresponding to the subject (10). A content substitution unit (47) substitutes the marked areas with alternate image content (42) according to the mask signal (43) to output modified camera images (41). An image selector unit (48) selects the alternate image content (42) amongst at least a first alternate image content (42a) when the subject is determined to be in a first condition within the scene and a second alternate image content (42b) when the subject is determined to be in a second condition within the scene. In examples, the first and second alternate image contents (42a, 42b) are selected based on a determined camera zoom value or a camera angle.