Research map / 3D vision / Human pose estimation
LA-Pose: Latent Action Pretraining Meets Pose Estimation
Zhengqing Wang, Saurabh Nair, Prajwal Chidananda, Pujith Kachana, Samuel Li, Matthew Brown, Yasutaka Furukawa
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| compress latent action to 50 dimensions instead of 1536 (ablation) | LA-Pose · Waymo | ATE-M: 1.94 -> 1.62 | Improved |
| compress latent action to 50 dimensions instead of 1536 (ablation) | LA-Pose · Waymo | AUC@5: 86.5 -> 85.4 | No clear effect |
| use larger 1536-D latent action (ablation) | LA-Pose · Waymo | pre-training loss @200k: 1.67 -> 1.15 | Improved |
| freeze pretrained inverse-dynamics encoder instead of fine-tuning (ablation) | LA-Pose · PandaSet (zero-shot) | ATE-M | Improved |
| freeze pretrained inverse-dynamics encoder instead of fine-tuning (ablation) | LA-Pose · Waymo | ATE-M | No clear effect |
| replace VGGT with LA-Pose at 1 fps | LA-Pose · Waymo (1.0 fps) | AUC@5: 74.6 -> 85.7 | Improved |
| replace VGGT with LA-Pose at 4 fps | LA-Pose · Waymo (4.0 fps) | AUC@5: 74.1 -> 93.4 | Improved |
| replace VGGT with LA-Pose on unseen PandaSet | LA-Pose · PandaSet (zero-shot) | AUC@5: 75.0 -> 86.3 | Improved |
| replace VGGT with LA-Pose on unseen PandaSet | LA-Pose · PandaSet (zero-shot) | ATE-S: 0.99 -> 1.13 | Hurt |
| replace VGGT with LA-Pose on Waymo | LA-Pose · Waymo | ATE-S: 1.43 -> 1.20 | Improved |
| lower frame rate from 4.0 to 1.0 fps | LA-Pose · Waymo | AUC@5: 93.4 -> 85.7 | Hurt |
| medium-curvature trajectories vs straight trajectories | LA-Pose · Waymo val | AUC@5: 94.50 -> 78.32 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- VGGT — VGGT: Visual Geometry Grounded Transformer (CVPR 2025) · better on 9 of 10 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Predictive Inverse Dynamics Models are Scalable Learners for Robotic Manipulation (ICLR 2025 · similarity 0.88)
- PlaySlot: Learning Inverse Latent Dynamics for Controllable Object-Centric Video Prediction and Planning (ICML 2025 · similarity 0.88)
- Latent Action Pretraining from Videos (ICLR 2025 · similarity 0.88)
- RotVLA: Rotational Latent Action for Vision-Language-Action Model (NeurIPS 2026 · similarity 0.88)
- World Action Verifier: Self-Improving World Models via Forward-Inverse Asymmetry (NeurIPS 2026 · similarity 0.87)
- DiT4DiT: Jointly Modeling Video Dynamics and Actions for Generalizable Robot Control (NeurIPS 2026 · similarity 0.87)
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.
