Density Forests in C++

Figure 1: Example results from the density forests paper.

Density forests can be understood as random forests for density estimations. Instead of splitting examples at each leaf, each leaf models a Gaussiand distribution. In this sense, a density forest can be seen as a special case of Gaussian mixture models using hard assignments instead of soft assignments. Figure 1 shows examples from the original paper. The following repository includes a C++ implementation of density forests, bundled with standard decision forests implemented by Tobias Pohlen. See this article for some results with this implementation.

Density Forests on GitHub
  • A. Criminisi, J. Shotton. Density Forests. In Decision Forests for Computer Vision and Medical Image Analysis, Springer London, 2013.