Abstract:This paper aims to tackle the practically very challenging problem of efficient and accurate hand pose estimation from single depth images. A dedicated two-step regression forest pipeline is proposed: given an input hand depth image, step one involves mainly estimation of 3D location and in-plane rotation of the hand using a pixel-wise regression forest. This is utilized in step two which delivers final hand estimation by a similar regression forest model based on the entire hand image patch. Moreover, our estimation is guided by internally executing a 3D hand kinematic chain model. For an unseen test image, the kinematic model parameters are estimated by a proposed dynamically weighted scheme. As a combined effect of these proposed building blocks, our approach is able to deliver more precise estimation of hand poses. In practice, our approach works at 15.6 frame-per-second (FPS) on an average laptop when implemented in CPU, which is further sped-up to 67.2 FPS when running on GPU. In addition, we introduce and make publicly available a data-glove annotated depth image dataset covering various hand shapes and gestures, which enables us conducting quantitative analyses on real-world hand images. The effectiveness of our approach is verified empirically on both synthetic and the annotated real-world datasets for hand pose estimation, as well as related applications including part-based labeling and gesture classification. In addition to empirical studies, the consistency property of our approach is also theoretically analyzed.

Model-based Deep Hand Pose Estimation

Dual Regression for Efficient Hand Pose Estimation

Model-based Hand Pose Estimation for Generalized Hand Shape with Appearance Normalization

Deep Predictive Neural Network: Unsupervised Learning for Hand Pose Estimation

A CNN Model for Real Time Hand Pose Estimation.

Context-Aware Deep Spatio-Temporal Network for Hand Pose Estimation from Depth Images

Estimate Hand Poses Efficiently from Single Depth Images

Deep Kinematic Pose Regression.

Deep Conditional Variational Estimation for Depth-Based Hand Poses

DeepHPS: End-to-end Estimation of 3D Hand Pose and Shape by Learning from Synthetic Depth

Learning Hand Latent Features for Unsupervised 3D Hand Pose Estimation

HMTNet:3D Hand Pose Estimation from Single Depth Image Based on Hand Morphological Topology

Learning a Deep Predictive Coding Network for a Semi-Supervised 3D-Hand Pose Estimation

A Survey of Deep Learning-based Hand Pose Estimation

Cross-Modal Deep Variational Hand Pose Estimation

Weakly Supervised 3D Hand Pose Estimation via Biomechanical Constraints

A Dual-Branch Self-Boosting Framework for Self-Supervised 3D Hand Pose Estimation

Hand Pose Estimation through Semi-Supervised and Weakly-Supervised Learning

RePose: Learning Deep Kinematic Priors for Fast Human Pose Estimation

Depth-Based 3D Hand Pose Estimation: from Current Achievements to Future Goals

Towards Good Practices for Deep 3D Hand Pose Estimation