first commit

.gitignore

@@ -0,0 +1,175 @@
+.vscode 
+
+light-field-distance
+
+data/abc
+data/shapenet
+data/custom
+experiments/
+eval_emd.py
+eval_metrics_others.py
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#pdm.lock
+#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
+#   in version control.
+#   https://pdm.fming.dev/#use-with-ide
+.pdm.toml
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/
+src_convonet/utils/libmcubes/mcubes.cpython-39-x86_64-linux-gnu.so
+src_convonet/utils/libmesh/triangle_hash.cpython-39-x86_64-linux-gnu.so
+src_convonet/utils/libmise/mise.cpython-39-x86_64-linux-gnu.so
+src_convonet/utils/libsimplify/simplify_mesh.cpython-39-x86_64-linux-gnu.so
+src_convonet/utils/libvoxelize/voxelize.cpython-39-x86_64-linux-gnu.so

LICENSE.txt

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2023 Yizhi Wang
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

README.md

@@ -0,0 +1,145 @@
+# ARO-Net
+
+This is the official Pytorch implementation of the paper:
+
+ARO-Net: Learning Implicit Fields from Anchored Radial Observations. CVPR. 2023.
+
+Yizhi Wang*, Zeyu Huang*, Ariel Shamir, Hui Huang, Hao Zhang, Ruizhen Hu.
+
+Paper: [arxiv](https://arxiv.org/abs/2212.10275)
+Homepage: [ARO-Net]()
+
+<img src='imgs/aro-net.jpg'/>
+
+<img src='imgs/aro_net_demo.jpg'/>
+
+## Setup
+
+To setup a conda environment, and build dependencies:
+```
+# create conda environment for ARO-Net
+conda create -n aro-net python=3.9 pytorch-gpu -c pytorch
+conda activate aro-net
+pip install trimesh open3d tensorboard Cython
+
+# install tools from ConvONet
+python setup.py build_ext --inplace
+
+# install LFD metrics
+git clone https://github.com/kacperkan/light-field-distance
+cd light-field-distance
+python setup.py install
+```
+
+## Dataset
+
+The dataset used in our experiments can be found in [OneDrive](https://1drv.ms/u/s!AkDQSKsmQQCghcNbst1PuHeb-obv7w?e=wlGLCK) and [Baidu Disk](https://pan.baidu.com/s/1M1UQHV2Wv1g3lemqErUFlg) (Password: 2vde).
+
+It contains ABC dataset (the first Chunk), and ShapeNet (Chairs and Airplanes).
+
+The layout of ``data`` dir is: 
+
+```
+ARO-Net
+├── data
+│   ├── abc
+    │   │   │──00_meshes
+    │   │   │──01_pcds
+    │   │   │──02_qry_pts
+    │   │   │──03_qry_dists
+    │   │   │──04_splits        
+    ├── shapenet
+    │   │   │──00_meshes
+    │   │   │   │──02691156
+    │   │   │   │──03001627      
+    │   │   │──01_pcds
+    │   │   │──02_qry_pts
+    │   │   │──03_qry_dists
+    │   │   │──04_splits            
+    ├── anchors
+```
+
+## Quick Testing
+
+We provide pre-trained models on ABC dataset (first chunk) and ShapeNet (chairs):
+
+- ABC: [OneDrive](https://1drv.ms/f/s!AkDQSKsmQQCghcNfVnUQWiaw5mY59Q?e=xa5cZ9) or [Baidu Disk](https://pan.baidu.com/s/1qiVKt7SvXIoKBfQJEv2Z1g) (Password: hcwk).
+- ShapeNet: [Onedrive](https://1drv.ms/f/s!AkDQSKsmQQCghcNggvV_2b0kuexCaw?e=gjAOHj) or [Baidu Disk](https://pan.baidu.com/s/14CzMY_Q8DF8xXZbfOVw9JA) (Password: x6cj).
+
+Put them under the folder `experiments`. For ShapeNet dataset, we trained ARO-Net on two kinds of query-occupancy ground-truth provied by [IM-Net](https://github.com/czq142857/IM-NET-pytorch) and [OCC-Net](https://github.com/autonomousvision/occupancy_networks), respectively.
+
+To test our pretrained models,
+```
+# ABC
+python reconstruct.py --name_exp pretrained_abc --name_ckpt aronet_shapenet_chairs_gt_imnet --name_dataset abc --use_dist_hit True --norm_coord False
+
+# ShapeNet Chair
+python reconstruct.py --name_exp pretrained_shapenet_chairs --name_ckpt aronet_shapenet_chairs_gt_occnet.ckpt --name_dataset shapenet --categories_test 03001627, --use_dist_hit False --mc_threshold 0.2
+
+# ShapeNet Airplane
+python reconstruct.py --name_exp pretrained_shapenet_chairs --name_ckpt aronet_shapenet_chairs_gt_occnet.ckpt --name_dataset shapenet --name_dataset 02691156, --use_dist_hit False
+```
+You can also modify `--n_pts_test` to set the input of points of objects, we pre-sampled `1024` and `2048` points from meshes for testing.
+
+## Training
+
+We use Fibonacci sampling to generate 48 anchors for our ARO-Net. Other anchor settings can generated with `gen_anc.py`.
+
+To train ARO-Net on ABC dataset or ShapeNet:
+```
+python train.py --name_exp base_model_chairs --name_dataset shapenet --categories_train 03001627,
+python train.py --name_exp base_model_abc --name_dataset abc --use_dist_hit True
+```
+It is recommended to set `use_dist_hit` to `True` when training on abc dataset (an auxiliary loss to predict anchor-query to surface distance) when training (it will bring some marginal performance gain). To use this auxiliary loss, first run `cal_hit_dist.py`.
+
+To train ARO-Net on single shape with data augmentation:
+```
+python train.py --name_exp base_model --name_dataset single --name_single fertility
+```
+
+Check all training options in `options.py`. You need one NVIDIA A100 (80G) to train ARO-Net under the configurations in `options.py`. You can set the `n_bs` and `n_qry` to fit to your GPU capacity. set `n_bs` to `4` and `n_qry` to `256` will cost ~20GB video memory.
+
+## Evaluation
+
+To reconstruct meshes on test sets:
+```
+# ABC
+python reconstruct.py --name_exp base_model_abc --name_ckpt 600_301101_xxx_xxx.ckpt --name_dataset abc --use_dist_hit True
+
+# ShapeNet Chair
+python reconstruct.py --name_exp base_model_chairs --name_ckpt 600_301101_xxx_xxx.ckpt --name_dataset shapenet --categories_test 03001627,
+
+# ShapeNet Airplane
+python reconstruct.py --name_exp base_model_chairs --name_ckpt 600_301101_xxx_xxx.ckpt --name_dataset shapenet --name_dataset 02691156,
+```
+
+To evalute HD, CD, and IoU:
+```
+# ABC
+python eval_metrics.py --name_exp base_model_abc --name_dataset abc
+
+# ShapeNet
+python eval_metrics.py --name_exp base_model_chairs --name_dataset shapenet --categories_test 03001627,
+python eval_metrics.py --name_exp base_model_chairs --name_dataset shapenet --categories_test 02691156,
+```
+
+To evaluate LDF:
+```
+# ABC
+python eval_lfd.py --name_exp 202203014_pretrained --name_dataset abc
+```
+We use [light-field-distance](https://github.com/kacperkan/light-field-distance) to compute LDF. The implementation of this library prevents us from computing this metric in parallel. It also requires an OpenGL context, so a pyhsical display is recommended. 
+
+
+## Acknowledgement
+
+[IM-Net](https://github.com/czq142857/IM-NET-pytorch)
+
+[Occ-Net](https://github.com/autonomousvision/occupancy_networks)
+
+[ConvONet](https://pengsongyou.github.io/conv_onet)
+
+[Points2Surf](https://github.com/ErlerPhilipp/points2surf)
+
+[light-field-distance](https://github.com/kacperkan/light-field-distance)
+