Research map / 3D vision / Gaussian splatting
Energy-GS: Image Energy-guided Pose Alignment Gaussian Splatting with redesigned pose gradient flow
Yu Gao, Lutong Su, Ruixiang Huang, Tianji Jiang, Jiadong Tang, Yufeng Yue, Yi Yang
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 |
|---|---|---|---|
| redesign pose gradient flow (fixed Gaussian positions, fixed tile support, no early densification) (ablation) | 3DGS with learnable poses · Synthetic ship | PSNR: 8.08 -> 12.38 | Improved |
| redesign pose gradient flow (ablation) | 3DGS with learnable poses · Synthetic ship | rotation error (deg): 8.572 -> 7.087 | Improved |
| add image energy control (progressive low-to-high energy supervision) on top of gradient redesign (ablation) | Energy-GS · Synthetic ship | PSNR: 12.38 -> 24.12 | Improved |
| add image energy control (ablation) | Energy-GS · Synthetic ship | ATE (m): 0.150 -> 0.011 | Improved |
| add image energy control (ablation) | Energy-GS · Synthetic ship | rotation error (deg): 7.087 -> 1.065 | Improved |
| add learnable camera poses to plain 3DGS with only RGB supervision (ablation) | 3DGS · Synthetic ship | PSNR | Hurt |
| add progressive alignment (energy control) after stabilizing pose gradients (ablation) | Energy-GS · Synthetic scenes | multi-shell failure mode | Improved |
| 1D signal alignment: redesigned gradient flow plus energy strategy versus original gradient flow (ablation) | 1D signal alignment toy task · 1D signal alignment | ATE: 0.1205 -> 0.0001 | Improved |
| Energy-GS vs BARF on synthetic scenes (rendering) | BARF · Synthetic dataset | PSNR: chair 28.35 -> 29.81 | No clear effect |
| Energy-GS vs BARF on synthetic scenes (pose) | BARF · Synthetic dataset | rotation error (deg): hotdog 0.653 -> 0.054 | Improved |
| Energy-GS vs 3R-GS on Mip-NeRF 360 | 3R-GS · Mip-NeRF 360 | PSNR / pose error: garden PSNR 25.23 vs 22.16; stump PSNR 18.87 vs 21.75 | Mixed |
| Energy-GS vs plain 3DGS with pose gradients | 3DGS · Synthetic chair | PSNR: 15.53 -> 29.81 | 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.
- BARF — BARF: Bundle-Adjusting Neural Radiance Fields (ICCV 2021) · better on 8 of 8 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
- COLMAP-Free 3D Gaussian Splatting (CVPR 2024 · similarity 0.89)
- VA-GS: Enhancing the Geometric Representation of Gaussian Splatting via View Alignment (NeurIPS 2025 · similarity 0.89)
- GS-CPR: Efficient Camera Pose Refinement via 3D Gaussian Splatting (ICLR 2025 · similarity 0.89)
- GS-SLAM: Dense Visual SLAM with 3D Gaussian Splatting (CVPR 2024 · similarity 0.89)
- KeyGS: A Keyframe-Centric Gaussian Splatting Method for Monocular Image Sequences (AAAI 2025 · similarity 0.89)
- MS-GS: Multi-Appearance Sparse-View 3D Gaussian Splatting in the Wild (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.
