PT Unknown
AU Wenjuan Gong
   Y.Huang
   Jordi Gonzalez
   Liang Wang
TI An Effective Solution to Double Counting Problem in Human Pose Estimation
PY 2015
DE Pose estimation; double counting problem; mix-ture of parts Model
AB The mixture of parts model has been successfully applied to solve the 2Dhuman pose estimation problem either as an explicitly trained body part modelor as latent variables for pedestrian detection. Even in the era of massiveapplications of deep learning techniques, the mixture of parts model is stilleffective in solving certain problems, especially in the case with limitednumbers of training samples. In this paper, we consider using the mixture ofparts model for pose estimation, wherein a tree structure is utilized forrepresenting relations between connected body parts. This strategy facilitatestraining and inferencing of the model but suffers from double countingproblems, where one detected body part is counted twice due to lack ofconstrains among unconnected body parts. To solve this problem, we propose ageneralized solution in which various part attributes are captured by multiplefeatures so as to avoid the double counted problem. Qualitative andquantitative experimental results on a public available dataset demonstrate theeffectiveness of our proposed method. An Effective Solution to Double Counting Problem in Human Pose Estimation – ResearchGate. Available from: http://www.researchgate.net/publication/271218491_An_Effective_Solution_to_Double_Counting_Problem_in_Human_Pose_Estimation [accessed Oct 22, 2015].
ER