Research map / 3D vision / Human pose estimation
BigMaQ: A Big Macaque Motion and Animation Dataset Bridging Image and 3D Pose Representations
Lucas Martini, Alexander Lappe, Anna Bognár, Rufin Vogels, Martin A. Giese
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 |
|---|---|---|---|
| add 3D-Rot pose vectors to ResNet50 features (ablation) | ResNet50 · BigMaQ500 | mAP: 34.3 -> 44.0 | Improved |
| add pose vectors to ViT-base-cls features (ablation) | ViT-base-cls · BigMaQ500 | mAP: 32.9 -> 44.0 | Improved |
| add pose vectors to TimeSformer features (ablation) | TimeSformer · BigMaQ500 | mAP: 31.9 -> 42.6 | Improved |
| add pose vectors to VideoPrism-base features (ablation) | VideoPrism-base · BigMaQ500 | mAP: 38.3 -> 43.8 | Improved |
| add pose vectors to DINOv2-base (patch tokens) features (ablation) | DINOv2-base · BigMaQ500 | mAP: 40.4 -> 41.4 | No clear effect |
| pose-only stream versus best visual-only model (ablation) | pose-only · BigMaQ500 | mAP: 43.5 vs 40.4 (DINOv2-base) | Improved |
| use 3D-Rot joint rotations instead of 2D keypoints (ablation) | pose representation · BigMaQ500 pose-only | mAP: 35.6 -> 43.5 | Improved |
| use 3D-Rot instead of 3D keypoints in pose-only stream (ablation) | pose representation · BigMaQ500 pose-only | mAP: 40.8 -> 43.5 | Improved |
| use 3D-Rot instead of 3D keypoints combined with VideoPrism (ablation) | VideoPrism-base · BigMaQ500 | mAP: 34.0 -> 43.8 | Improved |
| subject-specific mesh with temporal loss versus MAMMAL template | BigMaQ tracking · single-subject actions | IoU: 0.714 -> 0.844 | Improved |
| subject-specific mesh versus MAMMAL on 3D alignment | BigMaQ tracking · single-subject actions | MPJPE (mm): 31.661 -> 26.907 | Improved |
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.
- AniMer+ — AniMer: Animal Pose and Shape Estimation Using Family Aware Transformer (CVPR 2025) · better on 1 of 1 shared result
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
- MBW: Multi-view Bootstrapping in the Wild (NeurIPS 2022 · similarity 0.89)
- Animal3D: A Comprehensive Dataset of 3D Animal Pose and Shape (ICCV 2023 · similarity 0.88)
- BEAST3D: animal behavioral analysis and neural encoding from multi-view video via Gaussian splatting (NeurIPS 2026 · similarity 0.88)
- PriVi: Towards a General-Purpose Video Model for Primate Behavior in the Wild (CVPR 2026 · similarity 0.88)
- ChimpACT: A Longitudinal Dataset for Understanding Chimpanzee Behaviors (NeurIPS 2023 · similarity 0.88)
- Pose Splatter: A 3D Gaussian Splatting Model for Quantifying Animal Pose and Appearance (NeurIPS 2025 · 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.
