Research map / 3D vision / 3D generation
GrowFields: Compositional 4D Neural Fields for Topology-Changing Plant Growth
Joaquin Gajardo, Michele Volpi, Marko Mihajlovic, Siyu Tang, Lukas Roth, Sergey Prokudin
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 organ latent code (ablation) | GrowFields · TrackPlant3D full plant (8 test sequences, mean) | Chamfer distance / EPE: 1.05/3.44 -> 0.86/1.47 | Improved |
| add per-organ canonicalisation (ablation) | GrowFields · TrackPlant3D full plant (8 test sequences, mean) | Chamfer distance / EPE: 0.96/4.23 -> 0.86/1.47 | Improved |
| use direct latent concatenation instead of FiLM conditioning (ablation) | GrowFields · TrackPlant3D full plant (8 test sequences, mean) | Chamfer distance / EPE: 0.96/2.47 -> 0.86/1.47 | Improved |
| add multi-step Chamfer loss (ablation) | GrowFields · TrackPlant3D full plant (8 test sequences, mean) | Chamfer distance / EPE: 0.91/1.89 -> 0.86/1.47 | Improved |
| add latent code regularisation L_code (ablation) | GrowFields · TrackPlant3D full plant (8 test sequences, mean) | Chamfer distance / EPE: 0.88/1.74 -> 0.86/1.47 | Improved |
| use shared latent-conditioned field instead of per-organ MLPs (ablation) | GrowFields · TrackPlant3D full plant (8 test sequences, mean) | Chamfer distance / EPE: 0.93/2.37 -> 0.86/1.47 | Improved |
| use shared latent-conditioned field instead of per-organ MLPs (ablation) | GrowFields · TrackPlant3D per-organ sequences (45 organ seqs, mean) | Chamfer distance / EPE: 0.94/2.44 -> 0.87/1.62 | Improved |
| replace ground-truth organ segmentation with automatic segmentation | GrowFields · TrackPlant3D full plant (8 test sequences, mean) | Chamfer distance / EPE: 0.86/1.47 -> 0.86/1.42 | No clear effect |
| use part-aware compositional field versus full-plant baselines | GrowFields · TrackPlant3D full plant (8 test sequences, mean) | Chamfer distance: 615.7 -> 0.86 | Improved |
| use GrowFields instead of COAP for extrapolation | GrowFields · TrackPlant3D extrapolated frames (mean) | Chamfer distance: 451.09 -> 149.63 | Improved |
| use GrowFields instead of per-organ DPF for interpolation | GrowFields · TrackPlant3D interpolated frames (full plant, mean) | Chamfer distance: 10.24 -> 10.29 | Hurt |
| add Gaussian noise sigma 0.01 to input points | GrowFields · TrackPlant3D full plant (8 test sequences, mean) | Chamfer distance: 0.86 -> 1.08 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- COAP — COAP: Compositional Articulated Occupancy of People (CVPR 2022) · better on 6 of 6 shared results
- NVFi — NVFi: Neural Velocity Fields for 3D Physics Learning from Dynamic Videos (NeurIPS 2023) · better on 2 of 2 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
- Learning to Infer Parameterized Representations of Plants from 3D Scans (CVPR 2026 · similarity 0.89)
- Geometry-Centered 3D Latent World Models for Growing Surfaces (NeurIPS 2026 · similarity 0.85)
- Neural Hierarchical Decomposition for Single Image Plant Modeling (CVPR 2025 · similarity 0.85)
- SVDTree: Semantic Voxel Diffusion for Single Image Tree Reconstruction (CVPR 2024 · similarity 0.84)
- Birth and Death of a Rose (CVPR 2025 · similarity 0.83)
- TRENDy: Temporal Regression of Effective Nonlinear Dynamics (ICLR 2025 · similarity 0.83)
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.
