Research map / 3D vision / Human pose estimation
WildProp: Visual Estimation of Wildlife Body Proportions at Scale
Mustafa Chasmai, Aaron Sun, Subhransu Maji
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 pose-aware retrieval over random sampling (ablation) | WildProp · AVONET | relative error (%): 216.1 -> 40.0 | Improved |
| add SAM background suppression (ablation) | WildProp · AVONET | relative error (%): 40.0 -> 16.1 | Improved |
| add RANSAC geometric filtering (ablation) | WildProp · AVONET | relative error (%): 16.1 -> 12.2 | Improved |
| add iterative refinement (ablation) | WildProp · AVONET | relative error (%): 12.2 -> 11.3 | Improved |
| use CLIP instead of DINOv3 for retrieval (ablation) | WildProp · AVONET | relative error (%): 11.3 -> 17.1 | Hurt |
| use DINOv2 patch retrieval instead of DINOv3 (ablation) | WildProp · AVONET | relative error (%): 11.3 -> 10.1 | Mixed |
| use CLS token instead of patch concatenation (ablation) | WildProp · AVONET | relative error (%): 11.3 -> 11.3 | No clear effect |
| use similarity-weighted mean in refinement (ablation) | WildProp · AVONET | relative error (%): 12.0 -> 11.3 | Improved |
| remove top-k keypoint filter restriction (ablation) | WildProp · AVONET | relative error (%): 11.3 -> 11.3 | No clear effect |
| use Stable Diffusion features instead of DINOv3 for matching (ablation) | WildProp · AVONET | relative error (%): 11.3 -> 25.2 | Hurt |
| concatenate Stable Diffusion with DINOv3 features (ablation) | WildProp · AVONET | relative error (%): 11.3 -> 11.2 | No clear effect |
| use DINOv2 instead of DINOv3 for keypoint matching (ablation) | WildProp · AVONET | relative error (%): 11.3 -> 13.5 | Hurt |
Closest papers by meaning
- WildDepth: A Multimodal Dataset for 3D Wildlife Perception and Depth Estimation (ECCV 2026 · similarity 0.90)
- Reconstructing Animals and the Wild (CVPR 2025 · similarity 0.86)
- Generating the Wild: Individual-Consistent Image-to-Video Generation for Wildlife (NeurIPS 2026 · similarity 0.85)
- BirdCollect: A Comprehensive Benchmark for Analyzing Dense Bird Flock Attributes (AAAI 2024 · similarity 0.84)
- Web-Scale Collection of Video Data for 4D Animal Reconstruction (NeurIPS 2025 · similarity 0.84)
- Learning the 3D Fauna of the Web (CVPR 2024 · similarity 0.84)
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.
