ICML 2025 accepted papers on 3D vision
40 of the 3,342 papers accepted at ICML 2025 (1.2%) are on 3D vision, compared with 1.1% at ICML 2024. They fall into 6 topics.
See ICML 2025 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest accepted papers by meaning and how crowded the spot has become.
Topics and papers
3D generation · 19 papers
- Compositional Flows for 3D Molecule and Synthesis Pathway Co-design (1 citation)
- Geometry Informed Tokenization of Molecules for Language Model Generation (1 citation)
- Flex3D: Feed-Forward 3D Generation with Flexible Reconstruction Model and Input View Curation
- From Thousands to Billions: 3D Visual Language Grounding via Render-Supervised Distillation from 2D VLMs
- How to Move Your Dragon: Text-to-Motion Synthesis for Large-Vocabulary Objects
- Identifiable Object Representations under Spatial Ambiguities
- Introducing 3D Representation for Dense Volume-to-Volume Translation via Score Fusion
- Learning Fused State Representations for Control from Multi-View Observations
- Learning the RoPEs: Better 2D and 3D Position Encodings with STRING
- Orient Anything: Learning Robust Object Orientation Estimation from Rendering 3D Models
- Pre-training Auto-regressive Robotic Models with 4D Representations
- SafeMap: Robust HD Map Construction from Incomplete Observations
and 7 more in the interactive map.
Gaussian splatting · 7 papers
- MiraGe: Editable 2D Images using Gaussian Splatting (1 citation)
- 3D-LMVIC: Learning-based Multi-View Image Compression with 3D Gaussian Geometric Priors
- Perceptual-GS: Scene-adaptive Perceptual Densification for Gaussian Splatting
- ReferSplat: Referring Segmentation in 3D Gaussian Splatting
- VTGaussian-SLAM: RGBD SLAM for Large Scale Scenes with Splatting View-Tied 3D Gaussians
- PF3plat: Pose-Free Feed-Forward 3D Gaussian Splatting for Novel View Synthesis
- Tackling View-Dependent Semantics in 3D Language Gaussian Splatting
Face and shape · 6 papers
- Geometry-Informed Neural Networks (2 citations)
- FlowDrag: 3D-aware Drag-based Image Editing with Mesh-guided Deformation Vector Flow Fields
- Geometric Algebra Planes: Convex Implicit Neural Volumes
- Symmetry-Robust 3D Orientation Estimation
- Thickness-aware E(3)-Equivariant 3D Mesh Neural Networks
- PEINR: A Physics-enhanced Implicit Neural Representation for High-Fidelity Flow Field Reconstruction
Human pose estimation · 5 papers
- ADHMR: Aligning Diffusion-based Human Mesh Recovery via Direct Preference Optimization
- CUPS: Improving Human Pose-Shape Estimators with Conformalized Deep Uncertainty
- Imitation Learning from a Single Temporally Misaligned Video
- Semantics-aware Test-time Adaptation for 3D Human Pose Estimation
- ExtPose: Robust and Coherent Pose Estimation by Extending ViTs