# JGR-P2O: Joint Graph Reasoning based Pixel-to-Offset Prediction Network for 3D Hand Pose Estimation from a Single Depth Image

arXiv: https://arxiv.org/abs/2007.04646 · HTML: https://mlsanity.com/paper/2007.04646/

## Key facts

- Cited by 20 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 9 Jul 2020.

## Built on by (most cited first, 20 of 20)

- [Towards Accurate Alignment in Real-time 3D Hand-Mesh Reconstruction](https://mlsanity.com/paper/2109.01723/) (2021)
- [Hand Image Understanding via Deep Multi-Task Learning](https://mlsanity.com/paper/2107.11646/) (2021)
- [HandFoldingNet: A 3D Hand Pose Estimation Network Using Multiscale-Feature Guided Folding of a 2D Hand Skeleton](https://mlsanity.com/paper/2108.05545/) (2021)
- [A2J-Transformer: Anchor-to-Joint Transformer Network for 3D Interacting Hand Pose Estimation from a Single RGB Image](https://mlsanity.com/paper/2304.03635/) (2023)
- [Complementing Event Streams and RGB Frames for Hand Mesh Reconstruction](https://mlsanity.com/paper/2403.07346/) (2024)
- [HandDiff: 3D Hand Pose Estimation with Diffusion on Image-Point Cloud](https://mlsanity.com/paper/2404.03159/) (2024)
- [TriHorn-Net: A Model for Accurate Depth-Based 3D Hand Pose Estimation](https://mlsanity.com/paper/2206.07117/) (2022)
- [Neural Voting Field for Camera-Space 3D Hand Pose Estimation](https://mlsanity.com/paper/2305.04328/) (2023)
- [Image-free Domain Generalization via CLIP for 3D Hand Pose Estimation](https://mlsanity.com/paper/2210.16788/) (2022)
- [V-Hands: Touchscreen-based Hand Tracking for Remote Whiteboard Interaction](https://mlsanity.com/paper/2409.13347/) (2024)
- [Pushing the Envelope for Depth-Based Semi-Supervised 3D Hand Pose Estimation with Consistency Training](https://mlsanity.com/paper/2303.15147/) (2023)
- [HandDAGT: A Denoising Adaptive Graph Transformer for 3D Hand Pose Estimation](https://mlsanity.com/paper/2407.20542/) (2024)
- [EvHand-FPV: Efficient Event-Based 3D Hand Tracking from First-Person View](https://mlsanity.com/paper/2509.13883/) (2025)
- [Occlusion-Invariant Rotation-Equivariant Semi-Supervised Depth Based Cross-View Gait Pose Estimation](https://mlsanity.com/paper/2109.01397/) (2021)
- [Mask2Hand: Learning to Predict the 3D Hand Pose and Shape from Shadow](https://mlsanity.com/paper/2205.15553/) (2022)
- [Learning Pixel-wise Continuous Depth Representation via Clustering for Depth Completion](https://mlsanity.com/paper/2402.13579/) (2024)
- [DOR3D-Net: Dense Ordinal Regression Network for 3D Hand Pose Estimation](https://mlsanity.com/paper/2403.13405/) (2024)
- [DF-Mamba: Deformable State Space Modeling for 3D Hand Pose Estimation in Interactions](https://mlsanity.com/paper/2512.02727/) (2025)
- [HandMCM: Multi-modal Point Cloud-based Correspondence State Space Model for 3D Hand Pose Estimation](https://mlsanity.com/paper/2602.01586/) (2026)
- [KAD-Net: Kinematics-Aware Decoupled Learning for Robust 3D Hand Pose Estimation from a Single Depth Image](https://mlsanity.com/paper/2609.12559/) (2026)

## Abstract (arXiv)

State-of-the-art single depth image-based 3D hand pose estimation methods are based on dense predictions, including voxel-to-voxel predictions, point-to-point regression, and pixel-wise estimations. Despite the good performance, those methods have a few issues in nature, such as the poor trade-off between accuracy and efficiency, and plain feature representation learning with local convolutions. In this paper, a novel pixel-wise prediction-based method is proposed to address the above issues. The key ideas are two-fold: a) explicitly modeling the dependencies among joints and the relations between the pixels and the joints for better local feature representation learning; b) unifying the dense pixel-wise offset predictions and direct joint regression for end-to-end training. Specifically, we first propose a graph convolutional network (GCN) based joint graph reasoning module to model the complex dependencies among joints and augment the representation capability of each pixel. Then we densely estimate all pixels' offsets to joints in both image plane and depth space and calculate the joints' positions by a weighted average over all pixels' predictions, totally discarding the complex postprocessing operations. The proposed model is implemented with an efficient 2D fully convolutional network (FCN) backbone and has only about 1.4M parameters. Extensive experiments on multiple 3D hand pose estimation benchmarks demonstrate that the proposed method achieves new state-of-the-art accuracy while running very efficiently with around a speed of 110fps on a single NVIDIA 1080Ti GPU.