Priscilla K. Wagner,  Guilherme O. Borges,  Renata C. B. Madeo,  Sarajane M. Peres. 

This paper presents a tool for building benchmark datasets to evaluate image-based gesture analysis systems, considering object segmentation and pattern recognition tasks. The main motivation in this project is to provide an alternative to allow the efective comparisons among several iniciatives in the development of this type of systems. The tool has as a differential funcionalities for preparing dataset with videos, frames and feature vectors organized according acquision context, background style, performers, gesture sequences and gesture type. Such features allow to create test cases to benchmarking several systems properties.

http://www.lbd.dcc.ufmg.br/colecoes/sbsi/2012/0058.pdf

Processando consulta.