Research map / 3D vision / Human pose estimation
PoseScript: 3D Human Poses from Natural Language
Ginger Delmas, Philippe Weinzaepfel, Thomas Lucas, Francesc Moreno-Noguer, Grégory Rogez
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 |
|---|---|---|---|
| pretrain on automatic captions then finetune on human captions (ablation) | text-pose retrieval model · PoseScript-H | mRecall: 12.4 -> 30.4 | Improved |
| pretrain on automatic captions then finetune on human captions (ablation) | text-pose retrieval model · PoseScript-H | text-to-pose R@10: 19.4 -> 48.0 | Improved |
| train on human captions only instead of automatic captions | text-pose retrieval model · PoseScript-H | mRecall: 7.6 -> 12.4 | Improved |
| test on human captions after training on automatic captions only | text-pose retrieval model · PoseScript-A to PoseScript-H | mRecall: 69.1 -> 7.6 | Hurt |
| add extra KL regularizer LKL(Np,N0) (ablation) | text-conditioned VAE · PoseScript-A | FID: 0.10 -> 0.08 | Improved |
| add extra KL regularizer LKL(Np,N0) (ablation) | text-conditioned VAE · PoseScript-A | mRecall R/G: 24.7 -> 29.2 | Improved |
| add extra KL regularizer LKL(Np,N0) (ablation) | text-conditioned VAE · PoseScript-A | mRecall G/R: 14.4 -> 17.3 | Improved |
| pretrain on automatic captions (ablation) | text-conditioned VAE · PoseScript-H | mRecall R/G: 4.8 -> 15.4 | Improved |
| pretrain on automatic captions (ablation) | text-conditioned VAE · PoseScript-H | mRecall G/R: 2.7 -> 16.2 | Improved |
| pretrain on automatic captions (ablation) | text-conditioned VAE · PoseScript-H | FID: 0.14 -> 0.11 | Improved |
| use text-conditional pose prior instead of generic VPoser prior | SMPLify fitting · single SMPLify example | qualitative | Improved |
Closest papers by meaning
- PoseScript: Linking 3D Human Poses and Natural Language (TPAMI 2024 · similarity 0.99)
- ChatPose: Chatting about 3D Human Pose (CVPR 2024 · similarity 0.90)
- UniPose: A Unified Multimodal Framework for Human Pose Comprehension, Generation and Editing (CVPR 2025 · similarity 0.88)
- PoseGen: Learning to Generate 3D Human Pose Dataset with NeRF (AAAI 2024 · similarity 0.87)
- FlexPose: Pose Distribution Adaptation with Limited Guidance (AAAI 2025 · similarity 0.87)
- PoseLLaVA: Pose Centric Multimodal LLM for Fine-Grained 3D Pose Manipulation (AAAI 2025 · 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.
