Research map / 3D vision / 3D generation
HOSNeRF: Dynamic Human-Object-Scene Neural Radiance Fields from a Single Video
Jia-Wei Liu, Yan-Pei Cao, Tianyuan Yang, Zhongcong Xu, Jussi Keppo, Ying Shan, Xiaohu Qie, Mike Zheng Shou
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 state embeddings to the base model (ablation) | HOSNeRF base · HOSNeRF dataset, Backpack | LPIPS: 0.271 -> 0.252 | Improved |
| add object bones to the base model (without state embeddings) (ablation) | HOSNeRF base · HOSNeRF dataset, Backpack | LPIPS: 0.271 -> 0.269 | No clear effect |
| add object bones on top of state embeddings (ablation) | HOSNeRF · HOSNeRF dataset, Suitcase | LPIPS: 0.390 -> 0.386 | Improved |
| add foreground mask separation (ablation) | HOSNeRF · HOSNeRF dataset, Backpack | LPIPS: 0.253 -> 0.245 | Improved |
| add optical flow and cycle consistency supervision (ablation) | HOSNeRF · HOSNeRF dataset, Backpack | LPIPS: 0.245 -> 0.243 | Improved |
| add state embeddings, object bones, mask, flow and cycle versus base on Lounge scene (ablation) | HOSNeRF · HOSNeRF dataset, Lounge | LPIPS: 0.243 -> 0.227 | Improved |
| use HOSNeRF instead of NeuMan on collected dataset | dynamic human-object scenes · HOSNeRF dataset, Backpack | LPIPS: 0.478 -> 0.243 | Improved |
| use HOSNeRF (without objects) instead of NeuMan on NeuMan dataset | dynamic human-scene · NeuMan dataset, Seattle | LPIPS: 0.26 -> 0.14 | Improved |
| use HOSNeRF instead of NeuMan on PSNR for Lounge and Citron scenes | dynamic human-object-scene · HOSNeRF Lounge / NeuMan Citron | PSNR: 28.40 -> 27.74 (Lounge); 24.71 -> 24.44 (Citron) | Hurt |
Later papers that compared with this method
1 later accepted paper reports HOSNeRF as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Free-Viewpoint Video of Outdoor Sports Using a Drone (ECCV 2024 · better on 6 of 9 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
- HumanNeRF: Efficiently Generated Human Radiance Field From Sparse Inputs (CVPR 2022 · similarity 0.90)
- H-NeRF: Neural Radiance Fields for Rendering and Temporal Reconstruction of Humans in Motion (NeurIPS 2021 · similarity 0.89)
- Neural Human Performer: Learning Generalizable Radiance Fields for Human Performance Rendering (NeurIPS 2021 · similarity 0.88)
- HSR: Holistic 3D Human-Scene Reconstruction from Monocular Videos (ECCV 2024 · similarity 0.88)
- Neural Capture of Animatable 3D Human from Monocular Video (ECCV 2022 · similarity 0.88)
- Gear-NeRF: Free-Viewpoint Rendering and Tracking with Motion-aware Spatio-Temporal Sampling (CVPR 2024 · 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.
