Research map / 3D vision / NeRF and radiance fields
GNeRF: GAN-Based Neural Radiance Field Without Posed Camera
Quan Meng, Anpei Chen, Haimin Luo, Minye Wu, Hao Su, Lan Xu, Xuming He, Jingyi Yu
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 adversarial loss (ablation) | GNeRF · Synthetic-NeRF Chair | PSNR: 19.31 -> 31.30 | Improved |
| add inversion network (ablation) | GNeRF · Synthetic-NeRF Chair | PSNR: 13.82 -> 31.30 | Improved |
| add photometric refinement phase (ablation) | GNeRF · Synthetic-NeRF Chair | PSNR: 20.60 -> 31.30 | Improved |
| add iterative AB optimization between phases (A, AB...AB, B vs A, B) (ablation) | GNeRF · Synthetic-NeRF Chair | PSNR: 29.23 -> 31.30 | Improved |
| add iterative AB optimization (rotation error) (ablation) | GNeRF · Synthetic-NeRF Chair | rotation error (deg): 0.592 -> 0.363 | Improved |
| replace COLMAP poses with GNeRF's estimated poses on a low-texture scene | NeRF · DTU Scan48 | PSNR: 6.718 -> 23.25 | Improved |
| replace COLMAP poses with GNeRF's poses on a texture-rich scene | NeRF · Synthetic-NeRF Chair | PSNR: 33.75 -> 31.30 | Hurt |
| widen camera radius sampling from 4 to [3,5] (ablation) | GNeRF · Synthetic-NeRF Chair | PSNR: 31.30 -> 29.68 | Hurt |
| widen elevation range from [0,90] to [-90,90] (ablation) | GNeRF · Synthetic-NeRF Chair | PSNR: 31.30 -> 23.08 | Hurt |
| sample lookat from a Gaussian instead of a fixed point (ablation) | GNeRF · Synthetic-NeRF Chair | PSNR: 31.30 -> 30.48 | No clear effect |
| use GNeRF unmasked instead of IDR masked, no camera init | IDR · DTU Scan104 | PSNR: 12.26 -> 25.70 | Improved |
Later papers that compared with this method
4 later accepted papers report GNeRF as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- 3D Congealing: 3D-Aware Image Alignment in the Wild (ECCV 2024 · better on 2 of 2 shared results)
- Pose-Free Neural Radiance Fields via Implicit Pose Regularization (ICCV 2023 · better on 14 of 14 shared results)
- LU-NeRF: Scene and Pose Estimation by Synchronizing Local Unposed NeRFs (ICCV 2023 · better on 3 of 6 shared results)
- Seeing a Rose in Five Thousand Ways (CVPR 2023 · better on 5 of 5 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
- pixelNeRF: Neural Radiance Fields From One or Few Images (CVPR 2021 · similarity 0.90)
- SPARF: Neural Radiance Fields From Sparse and Noisy Poses (CVPR 2023 · similarity 0.90)
- Shape, Pose, and Appearance From a Single Image via Bootstrapped Radiance Field Inversion (CVPR 2023 · similarity 0.90)
- AltNeRF: Learning Robust Neural Radiance Field via Alternating Depth-Pose Optimization (AAAI 2024 · similarity 0.88)
- CF-NeRF: Camera Parameter Free Neural Radiance Fields with Incremental Learning (AAAI 2024 · similarity 0.88)
- ColNeRF: Collaboration for Generalizable Sparse Input Neural Radiance Field (AAAI 2024 · 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.
