Research map / 3D vision / Human pose estimation
VIBE: Video Inference for Human Body Pose and Shape 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 |
|---|---|---|---|
| Use VIBE's temporal generator and AMASS-trained motion discriminator (ablation) | SPIN-based pose and shape estimation · 3DPW | PA-MPJPE: 56.9 -> 51.9 | Helped |
| Add an AMASS-trained motion discriminator (ablation) | HMR-based temporal generator · 3DPW | PA-MPJPE: 75.8 -> 72.4 | Helped |
| Add an AMASS-trained motion discriminator (ablation) | SPIN-based temporal generator · 3DPW | PVE: 109.5 -> 99.1 | Helped |
| Replace the motion discriminator with the MPoser prior (ablation) | SPIN-based temporal generator · 3DPW | PA-MPJPE: 51.9 -> 54.1 | Hurt |
| Replace dynamic self-attention with static average-and-max concatenation (ablation) | VIBE motion discriminator · 3DPW | PA-MPJPE: 51.9 -> 53.7 | Hurt |
| Use 2 attention MLP layers with 512 nodes instead of 1024 (ablation) | VIBE motion discriminator · 3DPW | PA-MPJPE: 51.9 -> 54.2 | Hurt |
| Use 3 attention MLP layers with 512 nodes (ablation) | VIBE motion discriminator · 3DPW | MPJPE: 82.9 -> 85.3 | Hurt |
| Use 3 attention MLP layers with 1024 nodes (ablation) | VIBE motion discriminator · 3DPW | PA-MPJPE: 51.9 -> 52.4 | Hurt |
| Increase the sequence length from the selected 16-frame setup to other tested lengths (ablation) | VIBE temporal generator · Training sequence length | 3D pose estimation performance | Hurt |
| Train VIBE with the 3DPW training set | VIBE · 3DPW | PA-MPJPE: 56.5 -> 51.9 | Helped |
| Use video-based VIBE rather than the strongest reported frame-based result | 3D pose estimation · 3DPW | PA-MPJPE: 59.2 -> 51.9 | Helped |
| Use video-based pose estimation rather than the strongest reported frame-based result | 3D pose estimation · MPI-INF-3DHP | PA-MPJPE: 67.5 -> 63.4 | Helped |
Closest papers by meaning
- Capturing Humans in Motion: Temporal-Attentive 3D Human Pose and Shape Estimation From Monocular Video (CVPR 2022 · similarity 0.90)
- Learning 3D Human Dynamics From Video (CVPR 2019 · similarity 0.89)
- Convolutional Mesh Regression for Single-Image Human Shape Reconstruction (CVPR 2019 · similarity 0.89)
- 3D Human Pose Estimation Using Spatio-Temporal Networks with Explicit Occlusion Training (AAAI 2020 · similarity 0.88)
- Capturing the Motion of Every Joint: 3D Human Pose and Shape Estimation with Independent Tokens (ICLR 2023 · similarity 0.88)
- BEDLAM2.0: Synthetic humans and cameras in motion (NeurIPS 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.