Abstract
A new and robust mapping approach is proposed entitled mapping forests (MFs) for computer vision applications based on regression transformations. Mapping forests relies on learning nonlinear mappings deduced from pairs of source and target training data, and improves the performance of mappings by enabling nonlinear transformations using forests. In contrast to previous approaches, it provides automatically selected mappings, which are naturally nonlinear. MF can provide accurate nonlinear transformations to compensate the gap of linear mappings or can generalize the nonlinear mappings with constraint kernels. In our experiments, we demonstrate that the proposed MF approach is not only on a par or better than linear mapping approaches and the state-of-the-art, but also is very time efficient, which makes it an attractive choice for real-time applications. We evaluated the efficiency and performance of the MF approach using the BU3DFE and Multi-PIE datasets for multi-view facial expression recognition application, and Set5, Set14 and SuperTex136 datasets for single image super-resolution application.
Highlights
The first ideas of decision forests and subsequently random forests (RFs) belongs to more than two decades ago
We have evaluated the performance of our mapping forests (MFs) approach on two well-known computer vision open problems, namely multiview facial expression recognition (MFER) and single image super-resolution (SISR)
A qualitative comparison is shown in Fig. 6, which shows that our MF approach is successful for image super-resolution problem too
Summary
The first ideas of decision forests and subsequently random forests (RFs) belongs to more than two decades ago Due to their fast processing speed, forests have been extensively used to solve computer vision problems especially those requiring real-time processing. An extensive study on decision forests and their applications is provided by Criminisi et al [4]; where they discussed models for classification, manifold, supervised and semi-supervised learning, regression, density estimation, etc., by means of decision forests. They discussed the advantages and disadvantages of decision forests and extended the idea of decision forests to solve continuous problems such as regression and density estimation. Choosing an appropriate nonlinear kernel is difficult because it needs behavioral systems analysis (BSA) of the problem, which is not always feasible for complicated problems such as those involving
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
