Abstract: An example system includes: a landmark detection engine to detect landmark positions of landmarks in images based on facial detection; an optical flow landmark engine to determine the landmark positions in the images based on optical flow of the landmarks between the images; a landmark difference engine to determine, for a landmark in a given image: a distance between a detected landmark position and an optical flow landmark position of the landmark; and a weighted landmark determination engine to determine, for a first and second image, a position for the landmark in the second image based on: a respective detected landmark position and a respective optical flow position of the landmark in the second image; and respective distances, determined with the landmark difference engine, between a first detected landmark position of the landmark in the first image and respective optical flow landmark positions for the first and second images.

Abstract: An example device is described for facilitating polygon localization. In various aspects, the device can comprise a processor. In various instances, the device can comprise a non-transitory machine-readable memory that can store machine-readable instructions. In various cases, the processor can execute the machine-readable instructions, which can cause the processor to localize a polygon depicted in an image, based on execution of a deep learning pipeline. In various aspects, the deep learning pipeline can comprise a circular-softmax block.

Abstract: Examples of methods for image enhancement are described. In some examples, a method includes segmenting an image into an object region and a background region. In some examples, the image has a first resolution. In some examples, the method includes generating, using a first machine learning model, an enhanced object region with a second resolution that is greater than the first resolution. In some examples, the first machine learning model has been trained based on object landmarks. In some examples, the method includes generating, using a second machine learning model, an enhanced background region with a third resolution that is greater than the first resolution. In some examples, the method includes combining the enhanced object region and the enhanced background region to produce an enhanced image.

Abstract: Examples of methods for fiducial location are described herein. In some examples, a method may include estimating a first location of an unidentified fiducial based on image coordinates and grid coordinates of a set of contiguous fiducials. In some examples, the method may include determining that a candidate fiducial is a fiducial based on comparing the first location with a second location of the candidate fiducial.

Abstract: Examples of methods for fiducial pattern detection are described herein. In some examples, a method may include detecting fiducial pattern subsets in image subsets of an image of an object. In some examples, the method may also include selecting a first image subset that includes a largest first fiducial pattern subset. In some examples, the method may further include extending the first fiducial pattern subset from the first image subset to a neighboring second image subset.

Abstract: An example system includes: a landmark detection engine to detect landmark positions of landmarks in images based on facial detection; an optical flow landmark engine to determine the landmark positions in the images based on optical flow of the landmarks between the images; a landmark difference engine to determine, for a landmark in a given image: a distance between a detected landmark position and an optical flow landmark position of the landmark; and a weighted landmark determination engine to determine, for a first and second image, a position for the landmark in the second image based on: a respective detected landmark position and a respective optical flow position of the landmark in the second image; and respective distances, determined with the landmark difference engine, between a first detected landmark position of the landmark in the first image and respective optical flow landmark positions for the first and second images.