{"id":18614395,"url":"https://github.com/showlab/DeVRF","last_synced_at":"2025-04-11T00:30:37.656Z","repository":{"id":46755972,"uuid":"498183180","full_name":"showlab/DeVRF","owner":"showlab","description":"The Pytorch implementation of \"DeVRF: Fast Deformable Voxel Radiance Fields for Dynamic Scenes\"","archived":false,"fork":false,"pushed_at":"2022-10-17T12:38:51.000Z","size":1055,"stargazers_count":181,"open_issues_count":3,"forks_count":12,"subscribers_count":8,"default_branch":"main","last_synced_at":"2025-04-01T17:24:45.696Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"other","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/showlab.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE.txt","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null}},"created_at":"2022-05-31T04:17:21.000Z","updated_at":"2024-12-03T06:15:04.000Z","dependencies_parsed_at":"2022-09-24T08:20:56.468Z","dependency_job_id":null,"html_url":"https://github.com/showlab/DeVRF","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/showlab%2FDeVRF","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/showlab%2FDeVRF/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/showlab%2FDeVRF/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/showlab%2FDeVRF/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/showlab","download_url":"https://codeload.github.com/showlab/DeVRF/tar.gz/refs/heads/main","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":248322212,"owners_count":21084333,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2024-11-07T03:25:55.966Z","updated_at":"2025-04-11T00:30:36.992Z","avatar_url":"https://github.com/showlab.png","language":"Python","funding_links":[],"categories":["Papers"],"sub_categories":["NeRF Related Tasks"],"readme":"\n  \n# DeVRF: Fast Deformable Voxel Radiance Fields for Dynamic Scenes\n\n  \n\n[Project page](https://jia-wei-liu.github.io/DeVRF) | [arXiv](https://arxiv.org/abs/2205.15723)\n\n  \n\n\u003e **TL;DR:** A novel representation and learning paradigm for dynamic radiance fields reconstruction -- 100x faster, no loss in dynamic novel view synthesis quality.\n\n  \n\n\u003cimg  src=\"/figures/DeVRF.png\"  alt=\"DeVRF\"  style=\"zoom:67%;\"  /\u003e\n\n  \n\n## 📢 News\n\n - [2022.10.10]  We release the first-version of DeVRF code and dataset!\n\n- [2022.9.15] DeVRF got accepted by [**NeurIPS 2022**](https://nips.cc/)!\n\n- [2022.6.1] We release the arXiv paper.\n\n  \n\n## 📝 Preparation\n\n### Installation\n```\ngit clone https://github.com/showlab/DeVRF.git\ncd DeVRF\npip install -r requirements.txt\n```\n[Pytorch](https://pytorch.org/) and [torch_scatter](https://github.com/rusty1s/pytorch_scatter) installation is machine dependent, please install the correct version for your machine.\n\n\n\n\u003cdetails\u003e\n\n\u003csummary\u003e Dependencies (click to expand) \u003c/summary\u003e\n\n \n\n- `PyTorch`, `numpy`, `torch_scatter`, `pytorch3d`: main computation.\n\n- `scipy`, `lpips`: SSIM and LPIPS evaluation.\n\n- `tqdm`: progress bar.\n\n- `mmcv`: config system.\n\n- `opencv-python`: image processing.\n\n- `imageio`, `imageio-ffmpeg`: images and videos I/O.\n\n- `Ninja`: to build the newly implemented torch extention just-in-time.\n\n- `einops`: torch tensor shaping with pretty api.\n\n\u003c/details\u003e\n\n### DeVRF dataset\n\nWe release all the synthetic and real-world DeVRF dataset on [link](https://drive.google.com/drive/folders/1IuYCTIcUJxPJs6fE9SKSlLPNGiWs3OaQ?usp=sharing). DeVRF dataset consists of 5 inward-facing synthetic scenes (lego|floating_robot|kuka|daisy|glove), 1 inward-facing real-world scene (flower_360), and 3 forward-facing real-world scenes (plant|rabbit|pig_toy). For each scene, we release the static data, dynamic data, and optical flow estimated using [RAFT](https://github.com/princeton-vl/RAFT). Please refer to the following data structure for an overview of DeVRF dataset.\n\n```\n    DeVRF dataset\n    ├── inward-facing\n    │   └── [lego|floating_robot|kuka|daisy|glove|flower_360]\n    │       ├── static    \n    │       │\t├── [train|val|test]\n    │       │\t└── transforms_[train|val|test].json\n    │       └── dynamic_4views    \n    │        \t├── [train|val|test]\n    │        \t├── transforms_[train|val|test].json   \n    │        \t├── train_flow              \n    │        \t└── train_flow_png  \n    │          \n    └── forward-facing\n        └── [plant|rabbit|pig_toy]\n            ├── static    \n            │\t├── [images|images_4|images_8]\n            │\t└── poses_bounds.npy\n            └── dynamic_4views    \n             \t├── bds.npy  \n             \t├── poses.npy                        \n             \t└── [view1|view2|view3|view4]\n\t                ├── [images|images_4|images_8]\n             \t \t├── images_4_flow\n             \t \t└── images_4_flow_png\n```\n\nWe additionally provide a light version of DeVRF dataset without optical flow on [link](https://drive.google.com/drive/folders/18-1aRhFd7Z9ugZCOAZ9ZmHcoc9eaRBcg?usp=sharing).\n\n \n## 🏋️‍️ Experiment\n\n### Training\n\nStage 1: Train the static model using static scene data.\nThe static model part is almost the same as [DirectVoxGO](https://github.com/sunset1995/DirectVoxGO). The main difference is that we add an accumulated transmittance loss to encourage a clean background for forward-facing scenes. Please refer to [DirectVoxGO](https://github.com/sunset1995/DirectVoxGO) for more details.\n\nNote: Please enlarge the world_bound_scale in config file to establish a larger bounding box for dynamic scene modelling in the second stage. For DeVRF dataset, the world_bound_scale parameter is set within [1.05, 2.0].\n\t\n```bash\n\n$ cd static_DirectVoxGO\n$ python run.py --config configs/inward-facing/lego.py --render_test\n\n```\n\nStage 2: Train the dynamic model using dynamic scene data and the trained static model.\n\n```bash\n\n$ cd ..\n$ python run.py --config configs/inward-facing/lego.py --render_test\n\n```\n\n### Evaluation\n\nTo only evaluate the testset `PSNR`, `SSIM`, and `LPIPS` of the trained `lego` without re-training, run:\n\n```bash\n\n$ python run.py --config configs/inward-facing/lego.py --render_only --render_test \\\n\n--eval_ssim --eval_lpips_vgg --eval_lpips_alex\n\n```\n\nUse `--eval_lpips_alex` or `--eval_lpips_vgg` to evaluate LPIPS with pre-trained Alex net or VGG net.\n\n### Render video\n\t\n```bash\n\n$ python run.py --config configs/inward-facing/lego.py --render_only --render_video\n\n```\n\n### Reproduction: all config files to reproduce our results.\n\n\u003cdetails\u003e\n\n\u003csummary\u003e (click to expand) \u003c/summary\u003e\n\n```bash\n$ ls configs/*\n\nconfigs/inward-facing:\nlego.py floating_robot.py kuka.py daisy.py glove.py flower_360.py\n\nconfigs/forward-facing:\nplant.py rabbit.py pig_toy.py\n\n```\n \n\u003c/details\u003e\n\n  \n\n## 🎓 Citation\n\n  \n\nIf you find our work helps, please cite our paper.\n\n  \n\n```bibtex\n\n@article{liu2022devrf,\n\ntitle={DeVRF: Fast Deformable Voxel Radiance Fields for Dynamic Scenes},\n\nauthor={Liu, Jia-Wei and Cao, Yan-Pei and Mao, Weijia and Zhang, Wenqiao and Zhang, David Junhao and Keppo, Jussi and Shan, Ying and Qie, Xiaohu and Shou, Mike Zheng},\n\njournal={arXiv preprint arXiv:2205.15723},\n\nyear={2022}\n\n}\n\n```\n\n  \n\n## ✉️ Contact\n\n  \n\nThis repo is maintained by [Jiawei Liu](https://jia-wei-liu.github.io/). Questions and discussions are welcome via jiawei.liu@u.nus.edu.\n\n\n  \n\n## 🙏 Acknowledgements\n\n  \n\nThis codebase is based on [DirectVoxGO](https://github.com/sunset1995/DirectVoxGO).\n\n  \n\n\n## LICENSE\n\n  \n\nGPL","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fshowlab%2FDeVRF","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fshowlab%2FDeVRF","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fshowlab%2FDeVRF/lists"}