Mateusz Kozinski

  • Citations Per Year
Learn More
We address the problem of parsing images of building facades. The goal is to segment images, assigning to the resulting regions semantic labels that correspond to the basic architectural elements. We assume a top-down parsing framework based on a 2D shape grammar that encodes a prior knowledge on the possible composition of facades. The algorithm explores(More)
Existing approaches to parsing images of objects featuring complex, non-hierarchical structure rely on exploration of a large search space combining the structure of the object and positions of its parts. The latter task requires randomized or greedy algorithms that do not produce repeatable results or strongly depend on the initial solution. To address the(More)
We present a new shape prior formalism for the segmentation of rectified facade images. It combines the simplicity of split grammars with unprecedented expressive power: the capability of encoding simultaneous alignment in two dimensions, facade occlusions and irregular boundaries between facade elements. We formulate the task of finding the most likely(More)
We propose a novel formulation for parsing facade images with user-defined shape prior. Contrary to other state-of-the-art methods, we do not explore the procedural space of shapes derived from a grammar. Instead we formulate parsing as a linear binary program which we solve using Dual Decomposition. The algorithm produces plausible approximations of(More)
We propose a method for semi-supervised training of structured-output neural networks. Inspired by the framework of Generative Adversarial Networks (GAN), we train a discriminator network to capture the notion of a ‘quality’ of network output. To this end, we leverage the qualitative difference between outputs obtained on the labelled training data and(More)
As stated in section 2.1 of the paper, for any grid pattern shape prior, G = (C,R,H,V) there exists an adjacency pattern AG = (SG , V G , HG) encoding the same set of shapes. The set of pixel classes of the adjacency pattern is SG = R× C. We denote the row-class component of a pixel class s = (rs, cs) by r(s) = rs and its column-class component by c(s) =(More)
  • 1