Contemporary face recognition algorithms rely on precise
localization of keypoints (corner of eye, nose etc.). Unfortunately,
finding keypoints reliably and accurately remains
a...
City environments often lack textured areas, contain
repetitive structures, strong lighting changes and therefore
are very difficult for standard 3D modeling pipelines.
We prese...
Given a single image of an arbitrary road, that may not
be well-paved, or have clearly delineated edges, or some
a priori known color or texture distribution, is it possible
for...
In this paper, we present a new framework of video object
segmentation, in which we formulate the task of extracting
prominent objects from a scene as the problem of hypergraph
...
We present a method for the detection of instances of an
object class, such as cars or pedestrians, in natural images.
Similarly to some previous works, this is accomplished via
...