Abstract: In various examples, feature values corresponding to a plurality of views are transformed into feature values of a shared orientation or perspective to generate a feature map—such as a Bird's-Eye-View (BEV), top-down, orthogonally projected, and/or other shared perspective feature map type. Feature values corresponding to a region of a view may be transformed into feature values using a neural network. The feature values may be assigned to bins of a grid and values assigned to at least one same bin may be combined to generate one or more feature values for the feature map. To assign the transformed features to the bins, one or more portions of a view may be projected into one or more bins using polynomial curves. Radial and/or angular bins may be used to represent the environment for the feature map.

Abstract: A method for tracking vessels in image data includes receiving a plurality of image frames and annotating a target vessel in the plurality of image frames. A plurality of tracking targets associated with the target vessel is selected based on a first image frame included in the plurality of image frames. For each respective image frame, a set of tracking points are determined according to a tracking algorithm which includes: selecting a plurality of landmark hypothesis points for each tracking target in the respective frame based on a comparison with a previous image frame, constructing a directed acyclic graph from the plurality of landmark hypothesis points, and solving the directed acyclic graph to yield a plurality of optimal landmarks for the respective image frame.

Abstract: A method for tracking vessels in image data includes receiving a plurality of image frames and annotating a target vessel in the plurality of image frames. A plurality of tracking targets associated with the target vessel is selected based on a first image frame included in the plurality of image frames. For each respective image frame, a set of tracking points are determined according to a tracking algorithm which includes: selecting a plurality of landmark hypothesis points for each tracking target in the respective frame based on a comparison with a previous image frame, constructing a directed acyclic graph from the plurality of landmark hypothesis points, and solving the directed acyclic graph to yield a plurality of optimal landmarks for the respective image frame.