Research map / 3D vision / Human pose estimation
Keypoint Transformer: Solving Joint Identification in Challenging Hands and Object Interactions for Accurate 3D Pose Estimation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Replace per-joint heatmaps with detected keypoints and attention-based association | Keypoint Transformer · InterHand2.6M | MPJPE (All, mm): 14.22 -> 12.78 | Helped |
| Use detected keypoint features rather than features from the whole image | Keypoint Transformer · InterHand2.6M | MPJPE (All, mm): 14.48 -> 12.78 | Helped |
| Add decoder cross-attention to the CNN+SA baseline | CNN+SA · InterHand2.6M | MPJPE (All, mm): 15.31 -> 14.48 | Helped |
| Add keypoint-joint association loss (ablation) | Keypoint Transformer · InterHand2.6M interacting hand images | MPJPE (mm): 18.91 -> 17.08 | Helped |
| Add keypoint-joint association loss (ablation) | Keypoint Transformer · InterHand2.6M interacting hand images | MRRPE (mm): 38.96 -> 33.14 | Helped |
| Use 3D joint locations instead of MANO joint angles (ablation) | Keypoint Transformer · InterHand2.6M, two hands | MPJPE (mm): 20.61 -> 17.08 | Helped |
| Use 2.5D joint locations instead of MANO joint angles (ablation) | Keypoint Transformer · InterHand2.6M, two hands | MPJPE (mm): 20.61 -> 17.69 | Helped |
| Use 3D rather than 2.5D joint representation (ablation) | Keypoint Transformer · InterHand2.6M, all hands | MPJPE (mm): 14.73 -> 14.76 | No clear effect |
| Add self-attention layers (ablation) | Keypoint Transformer · InterHand2.6M, two hands | MPJPE (mm): 16.93 -> 15.44 | Helped |
| Increase cross-attention layers from 1 to 6 (ablation) | Keypoint Transformer · InterHand2.6M, two hands | MPJPE (mm): 15.44 -> 15.33 | No clear effect |
| Use ResNet-50 rather than ResNet-18 backbone (ablation) | Keypoint Transformer · InterHand2.6M, two hands | MPJPE (mm): 16.78 -> 15.32 | Helped |
| Apply Keypoint Transformer to HO-3D hand-object pose estimation | Keypoint Transformer · HO-3D | Joint Error: 3.04 -> 2.57 | Helped |
Closest papers by meaning
- A2J-Transformer: Anchor-to-Joint Transformer Network for 3D Interacting Hand Pose Estimation From a Single RGB Image (CVPR 2023 · similarity 0.90)
- DeepHand: Robust Hand Pose Estimation by Completing a Matrix Imputed With Deep Features (CVPR 2016 · similarity 0.90)
- Depth-Based 3D Hand Pose Estimation: From Current Achievements to Future Goals (CVPR 2018 · similarity 0.90)
- Robust 3D Hand Pose Estimation in Single Depth Images: From Single-View CNN to Multi-View CNNs (CVPR 2016 · similarity 0.90)
- 3D Hand Shape and Pose From Images in the Wild (CVPR 2019 · similarity 0.89)
- Reconstructing Hands in 3D with Transformers (CVPR 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.