Research map / 3D vision / Depth and stereo
RePer-360: Releasing Perspective Priors for 360° Depth Estimation via Self-Modulation
Cheng Guan, Chunyu Lin, Zhijie Shen, Junsong Zhang, Jiyuan Wang
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 ECCLoss cubemap consistency loss (ablation) | RePer-360 · Matterport3D (in-domain) | Abs Rel: 0.0716 -> 0.0691 | Improved |
| add SCAdaLN-Zero modulation instead of cross-attention fusion (ablation) | RePer-360 · Matterport3D (in-domain) | Abs Rel: 0.1168 -> 0.0691 | Improved |
| use geometry-aligned guidance instead of ERP only (ablation) | RePer-360 · Matterport3D (in-domain) | Abs Rel: 0.0698 -> 0.0691 | Improved |
| use geometry-aligned guidance instead of CP only (ablation) | RePer-360 · Matterport3D (in-domain) | RMSE: 0.3227 -> 0.3164 | Improved |
| use GAG instead of UniFuse CEE fusion (ablation) | RePer-360 · Matterport3D (in-domain) | Abs Rel: 0.0731 -> 0.0691 | Improved |
| replace panoramic attention with standard self-attention (ablation) | RePer-360 · Matterport3D (in-domain) | Abs Rel: 0.0691 -> 0.0725 | No clear effect |
| modulate odd layers in addition to output layers (ablation) | RePer-360 · Matterport3D (in-domain) | Abs Rel: 0.0750 -> 0.0691 | Improved |
| modulate all layers instead of Out.+Odd (ablation) | RePer-360 · Matterport3D (in-domain) | Abs Rel: 0.0691 -> 0.0733 | Hurt |
| use ECCLoss instead of EPNL (ablation) | RePer-360 · Matterport3D (in-domain) | Abs Rel: 0.0731 -> 0.0691 | Improved |
| add EPNL loss (ablation) | RePer-360 · Matterport3D (in-domain) | Abs Rel: 0.0716 -> 0.0731 | Mixed |
| initialize from DINOv2 instead of DAMv2 (ablation) | RePer-360 · Matterport3D (in-domain) | Abs Rel: 0.0691 -> 0.0808 | Hurt |
| RePer-360 vs PanDA-L under zero-shot | RePer-360 · Stanford2D3D (zero-shot) | Abs Rel: 0.1092 -> 0.0630 | Improved |
Closest papers by meaning
- DepthMaster: Unified Monocular Depth Estimation for Perspective and Panoramic Images (NeurIPS 2026 · similarity 0.91)
- Unified Panoramic Geometry Estimation via Multi-View Foundation Models (NeurIPS 2026 · similarity 0.89)
- Splatter-360: Generalizable 360 Gaussian Splatting for Wide-baseline Panoramic Images (CVPR 2025 · similarity 0.89)
- Depth Anywhere: Enhancing 360 Monocular Depth Estimation via Perspective Distillation and Unlabeled Data Augmentation (NeurIPS 2024 · similarity 0.87)
- RPG360: Robust 360 Depth Estimation with Perspective Foundation Models and Graph Optimization (NeurIPS 2025 · similarity 0.87)
- SharpDepth: Sharpening Metric Depth Predictions Using Diffusion Distillation (CVPR 2025 · similarity 0.87)
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.
