Collection of Dockerfiles for existing projects/papers I tested — useful as a starting point for your own experiments. Adapt the Dockerfiles and running scripts to your needs.
Each project/paper implementation consists of 3 files:
- Dockerfile
- build.sh
- run.sh
- README
and optional additional files (such as python requirements.txt, conda environemnts.yml, etc.).
The idea is to build the docker image from the Dockerfile using build.sh and run the docker container using run.sh. For a project <SELECTED-PROJECT> run the following:
cd dockers/<SELECTED-PROJECT>
sh build.sh
sh run.sh CODE-PATH DATA-PATH
where CODE-PATH is the path to the code from the original GitHub repository, and DATA-PATH is the optional path to the datasets location you want to use in the docker container. For more instructions read the dockers/<SELECTED-PROJECT>/README_<SELECTED-PROJECT>.md.
The implmemented projects/papers are listed in the table below. The files for a selected project are located in dockers/<Location>, where <Location> is listed in the table.
| Reference | Paper | Topic | Link | GitHub | Location |
|---|---|---|---|---|---|
| [1] | iNeRF: Inverting Neural Radiance Fields for Pose Estimation | NERFs | link | repo | iNeRF-public |
| [2] | SC^2-PCR: A Second Order Spatial Compatibility for Efficient and Robust Point Cloud Registration | 3D Registration | link | repo | SC2-PCR |
| [3] | Humans in 4D: Reconstructing and Tracking Humans with Transformers | SMPL fitting | link | repo | 4D-Humans |
| [5] | SpinNet: Learning a General Surface Descriptor for 3D Point Cloud Registration | 3D Registration | link | repo | SpinNet |
| [6] | Geometric Transformer for Fast and Robust Point Cloud Registration | 3D Registration | link | repo | GeoTransformer |
| [17] | Learning to Dress 3D People in Generative Clothing | SMPL clothing | link | repo | CAPE |
| [19] | Self-Calibrating Neural Radiance Fields | NERFs | link | repo | SCNeRF |
- Add GeDi [4] docker
- Add PointDSC [7] docker
- Add YOHO [8] docker
- Add RegTR [9] docker
- Add DIP [10] docker
- Add LVD [11] docker
- Add HierProb3D [12] docker
- Add VoteHMR [13] docker
- Add Unsupervised3DHuman [14] docker
- Add 3D-CODED [15] docker
- Add ALIKE [16] docker
- Add SMPLR [18] docker
- Add SC2-PCR [2] docker
- Add 4D-Humans [3] docker
- Add SpinNet [5] docker
- Add GeoTransformer [6] docker
- Add CAPE [17] docker
- Add SCNeRF [19] docker
[1] Yen-Chen et al.: iNeRF: Inverting Neural Radiance Fields for Pose Estimation, IROS 2021
[2] Chen et al.: SC^2-PCR: A Second Order Spatial Compatibility for Efficient and Robust Point Cloud Registration, CVPR 2022
[3] Goel et al.: Humans in 4D: Reconstructing and Tracking Humans with Transformers, arXiv preprint
[4] Poiesi et al.: Learning general and distinctive 3D local deep descriptors for point cloud registration, PAMI 2022
[5] Sheng et al.: SpinNet: Learning a General Surface Descriptor for 3D Point Cloud Registration, CVPR 2021
[6] Qin et al: Geometric Transformer for Fast and Robust Point Cloud Registration, CVPR 2022
[7] Bai et al: PointDSC: Robust Point Cloud Registration using Deep Spatial Consistency, CVPR 2021
[8] Wang et al: You only hypothesize once: Point cloud registration with rotation-equivariant descriptors, ACM MM 2022
[9] Yew et al: REGTR: End-to-end Point Cloud Correspondences with Transformers, CVPR 2022
[10] Poiesi et al.: Distinctive {3D} local deep descriptors, ICPR 2021
[11] Corona et al: Learned Vertex Descent: A New Direction for 3D Human Model Fitting, ECCV 2022
[12] Sengupta et al.: Hierarchical Kinematic Probability Distributions for 3D Human Shape and Pose Estimation from Images in the Wild, ICCV 2021
[13] Liu et al.: VoteHMR: Occlusion-Aware Voting Network for Robust 3D Human Mesh Recovery from Partial Point Clouds, ACM MM 2021
[14] Zuo et al.: Self-supervised 3D Human Mesh Recovery from Noisy Point Clouds, arxiv preprint
[15] Groueix et al.: 3D-CODED : 3D Correspondences by Deep Deformation, ECCV 2018
[16] Zhao et al.: ALIKE: Accurate and Lightweight Keypoint Detection and Descriptor Extraction, ToM 2022
[17] Ma et al.: Learning to Dress 3D People in Generative Clothing, CVPR 2020
[18] Madadi et al.: SMPLR: Deep learning based SMPL reverse for 3D human pose and shape recovery, Pattern Recognition 2020
[19] Jeong et al.: Self-Calibrating Neural Radiance Fields, ICCV 2021