Relevant Feature Selection for Human Pose Estimation and Localization in Cluttered Images

[摘要] We address the problem of estimating human body pose from a single image with cluttered background. We train multiple local linear regressors for estimating the 3D pose from a feature vector of gradient orientation histograms. Each linear regressor is capable of selecting relevant components of the feature vector depending on the poses by training it on a pose cluster which is a subset of training samples with similar poses.For discriminating the pose clusters, we use kernel Support Vector Machines with pose-dependent feature selection. Human detection is also capable with these SVMs, which are trained on synthetic data and are tested on real data since pose information is required for training. Quantitative experiments show that the effectiveness of pose-dependent feature selection to both human detection and pose estimation.

[发布机构] UCLA Henry Samueli School of Engineering and Applied Science

[学科分类] 计算机科学（综合）