Research map / 3D vision / Depth and stereo
PanoVGGT: Feed-Forward 3D Reconstruction from Panoramic Imagery
Yijing Guo, Mengjun Chao, Luo Wang, Tianyang Zhao, Haizhao Dai, Yingliang Zhang, Jingyi Yu, Yujiao Shi
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 PanoCity training data to panoramic pi3 (ablation) | pi3* · Matterport3D (3 views) | Abs Rel: 0.194 -> 0.105 | Improved |
| add tri-axis SO(3) rotation augmentation (ablation) | pi3* + PanoCity · Matterport3D (3 views) | AUC@30: 0.256 -> 0.285 | Improved |
| add tri-axis SO(3) rotation augmentation (ablation) | pi3* + PanoCity · Matterport3D (3 views) | mean translation error (deg): 31.52 -> 35.04 | Hurt |
| add global point fusion head (ablation) | PanoVGGT ablation chain · Matterport3D (3 views) | AUC@30: 0.285 -> 0.380 | Improved |
| add spherical sin-cos positional embedding (ablation) | PanoVGGT ablation chain · Matterport3D (3 views) | AUC@30: 0.380 -> 0.427 | Improved |
| retrain pi3 on panoramic data (pi3*) vs original pi3 | pi3 · Matterport3D pose | AUC@30: 0.047 -> 0.305 | Improved |
| use native panoramic PanoVGGT instead of pinhole pi3 via dodecahedral projection | pi3 · PanoCity point cloud | Overall mean: 8.222 -> 0.737 | Improved |
| use multi-view instead of monocular input | PanoVGGT · Stanford2D3D depth | Abs Rel: 0.0711 -> 0.0778 | Mixed |
| unified PanoVGGT vs per-dataset depth-only models | PanoVGGT · Structured3D depth | Abs Rel: 0.0273 (EGFormer) vs 0.0438 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- EGFormer — EGformer: Equirectangular Geometry-biased Transformer for 360 Depth Estimation (ICCV 2023) · better on 2 of 3 shared results
- Bifusev2 — BiFuse++: Self-Supervised and Efficient Bi-Projection Fusion for $360^{\circ }$ Depth Estimation (TPAMI 2022) · better on 2 of 2 shared results
- VGGT — VGGT: Visual Geometry Grounded 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
- Splatter-360: Generalizable 360 Gaussian Splatting for Wide-baseline Panoramic Images (CVPR 2025 · similarity 0.89)
- Unified Panoramic Geometry Estimation via Multi-View Foundation Models (NeurIPS 2026 · similarity 0.89)
- DepthMaster: Unified Monocular Depth Estimation for Perspective and Panoramic Images (NeurIPS 2026 · similarity 0.88)
- PanoWorld: Geometry-Consistent Panoramic Video World Modeling (NeurIPS 2026 · similarity 0.88)
- PanoPose: Self-supervised Relative Pose Estimation for Panoramic Images (CVPR 2024 · similarity 0.88)
- PanDA: Towards Panoramic Depth Anything with Unlabeled Panoramas and Mobius Spatial Augmentation (CVPR 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.
