| CPC G06V 10/764 (2022.01) [G06V 10/774 (2022.01); G06V 10/82 (2022.01); G06V 20/13 (2022.01)] | 21 Claims |
|
1. A computer-implemented method of classifying objects in an image of a surface of the Earth, the method comprising:
receiving image data associated with the image from a spaceborne or airborne detector;
receiving incidence angle data, wherein the incidence angle data is indicative of an incidence angle from which the image data is collected by the spaceborne or airborne detector;
using a machine learning model to classify one or more objects within the image as belonging to one of one or more categories based on the incidence angle data and values of one or more parameters of the image data, wherein the machine learning model is trained on a training dataset generated from training images and associated incidence angles; and
outputting the image showing classifications of one or more objects determined by the machine learning model,
wherein the training dataset is formed by:
receiving image data associated with a plurality of training images generated by a spaceborne or airborne detector and associated incidence angle for each image;
generating a training data patch for each training image by:
augmenting the associated incidence angle data to generate angle range data, wherein the angle range data is indicative of a range of angles that the machine learning model is trained to recognise as being similar to the incidence angle indicated by the incidence angle data; and
concatenating the angle range data to the associated image data; and
compiling the training data patches to generate the training dataset.
|