Computer aided traffic enforcement using dense correspondence estimation with multi-level metric learning and hierarchical matching

A system for detecting traffic scenarios, including:

• an image capturing device which captures two or more images of an area of a traffic environment with each image having a different view of vehicles and a road in the traffic environment;
• a hierarchical feature extractor configured for:
• concurrently extracting first features and second features at multiple neural network layers from each of the two or more images of the area, with the first features and the second features including geometric features and semantic features,
• estimating correspondences between the first features and the second features including using hierarchical fusing of geometric correspondences and semantic correspondences for each of the two or more images, and
• refining the estimated correspondences with correspondences between the geometric features of each of the two or more images to generate refined correspondence estimates;
• a traffic localization module for using the refined correspondence estimates to determine relative locations of elements and objects in the environment in three dimensions to automatically determine a traffic scenario according to the locations of vehicles according to a 3-dimensional (3D) model of a terrain generated based on the refined correspondence estimates;
• and
• a notification device configured to generate a notification of the traffic scenario based on the determined relative locations of elements and objects in the generated 3D model of the terrain.