RT Journal Article
T1 UPGMpp: a Software Library for Contextual Object Recognition
A1 Ruiz-Sarmiento, José Raúl
A1 Galindo-Andrades, Cipriano
A1 González-Jiménez, Antonio Javier
K1 Ingeniería de sistemas
AB Object recognition is a cornerstone task towards the sceneunderstanding problem. Recent works in the field boost their perfor-mance by incorporating contextual information to the traditional useof the objects’ geometry and/or appearance. These contextual cues areusually modeled through Conditional Random Fields (CRFs), a partic-ular type of undirected Probabilistic Graphical Model (PGM), and areexploited by means of probabilistic inference methods. In this work wepresent the Undirected Probabilistic Graphical Models in C++ library(UPGMpp), an open source solution for representing, training, and per-forming inference over undirected PGMs in general, and CRFs in par-ticular. The UPGMpp library supposes a reliable and comprehensiveworkbench for recognition systems exploiting contextual information, in-cluding a variety of inference methods based on local search, graph cuts,and message passing approaches. This paper illustrates the virtues of thelibrary, i.e. it is efficient, comprehensive, versatile, and easy to use, bypresenting a use-case applied to the object recognition problem in homescenes from the challenging NYU2 dataset.
YR 2015
FD 2015-09-07
LK http://hdl.handle.net/10630/10221
UL http://hdl.handle.net/10630/10221
LA eng
NO Universidad de Málaga. Campus de Excelencia Internacional Andalucía Tech.         Spanish grant program FPU-MICINN 2010and the Spanish projects “TAROTH: New developments toward a robot athome” (Ref. DPI2011-25483) and “PROMOVE: Advances in mobile roboticsfor promoting independent life of elders” (Ref. DPI2014-55826-R)
DS RIUMA. Repositorio Institucional de la Universidad de Málaga
RD 27 ene 2026