Implementation of CVPR'21: RfD-Net: Point Scene Understanding by Semantic Instance Reconstruction
RfD-Net: Point Scene Understanding by Semantic Instance Reconstruction
Yinyu Nie, Ji Hou, Xiaoguang Han, Matthias Nießner
In CVPR, 2021.
From an incomplete point cloud of a 3D scene (left), our method learns to jointly understand the 3D objects and reconstruct instance meshes as the output (right).
This implementation uses Python 3.6, Pytorch1.7.1, cudatoolkit 11.0. We recommend to use conda to deploy the environment.
conda env create -f environment.yml
conda activate rfdnet
pip install -r requirements.txt
Next, compile the external libraries by
python setup.py build_ext --inplace
Install PointNet++ by
export CUDA_HOME=/usr/local/cuda-X.X # replace cuda-X.X with your cuda version.
cd external/pointnet2_ops_lib
pip install .
The pretrained model can be downloaded here. Put the pretrained model in the directory as below
out/pretrained_models/pretrained_weight.pth
A demo is illustrated below to see how our method works.
cd RfDNet
python main.py --config configs/config_files/ISCNet_test.yaml --mode demo --demo_path demo/inputs/scene0549_00.off
VTK is used here to visualize the 3D scenes. The outputs will be saved under 'demo/outputs'. You can also play with your toy with this script.
If everything goes smooth, there will be a GUI window popped up and you can interact with the scene as below.
You can also use the offscreen
mode by setting offline=True
in demo.py
to render the 3D scene.
The rendered image will be saved in demo/outputs/some_scene_id/pred.png
.
In our paper, we use the input point cloud from the ScanNet dataset, and the annotated instance CAD models from the Scan2CAD dataset. Scan2CAD aligns the object CAD models from ShapeNetCore.v2 to each object in ScanNet, and we use these aligned CAD models as the ground-truth.
You can either directly download the processed samples [link] to the directory below (recommended)
datasets/scannet/processed_data/
or
datasets/scannet/scans
datasets/scannet/scan2cad_download_link
cd RfDNet
python utils/scannet/gen_scannet_w_orientation.py
You can either directly download the processed data [link] and extract them to datasets/ShapeNetv2_data/
as below
datasets/ShapeNetv2_data/point
datasets/ShapeNetv2_data/pointcloud
datasets/ShapeNetv2_data/voxel
datasets/ShapeNetv2_data/watertight_scaled_simplified
or
Download ShapeNetCore.v2 to the path below
datasets/ShapeNetCore.v2
Process ShapeNet models into watertight meshes by the following.
python utils/shapenet/1_fuse_shapenetv2.py
If it does not work, please delete the ./build
and .so
file in external/librender/
and recompile the pyrender by
cd RfDNet/external/librender
rm -rf ./build ./*.so
python setup.py build_ext --inplace
Sample points on ShapeNet models for training (similar to Occupancy Networks).
python utils/shapenet/2_sample_mesh.py --resize --packbits --float16
There are usually 100K+ points per object mesh. We simplify them to speed up our testing and visualization by
python utils/shapenet/3_simplify_fusion.py --in_dir datasets/ShapeNetv2_data/watertight_scaled --out_dir datasets/ShapeNetv2_data/watertight_scaled_simplified
After preprocessed the data, you can run the visualization script below to check if they are generated correctly.
Visualize ScanNet+Scan2CAD+ShapeNet samples by
python utils/scannet/visualization/vis_gt.py
A VTK window will be popped up like below.
We use the configuration file (see 'configs/config_files/****.yaml') to fully control the training/testing/generating process.
You can check a template at configs/config_files/ISCNet.yaml
.
We firstly pretrain our detection module and completion module followed by a joint refining. You can follow the process below.
Pretrain the detection module by
python main.py --config configs/config_files/ISCNet_detection.yaml --mode train
It will save the detection module weight at
out/iscnet/a_folder_with_detection_module/model_best.pth
Copy the weight path of detection module (see 1.) into configs/config_files/ISCNet_completion.yaml
as
weight: ['out/iscnet/a_folder_with_detection_module/model_best.pth']
Then pretrain the completion module by
python main.py --config configs/config_files/ISCNet_completion.yaml --mode train
It will save the completion module weight at
out/iscnet/a_folder_with_completion_module/model_best.pth
Copy the weight path of completion module (see 2.) into configs/config_files/ISCNet.yaml
as
weight: ['out/iscnet/a_folder_with_completion_module/model_best.pth']
Then jointly finetune RfD-Net by
python main.py --config configs/config_files/ISCNet.yaml --mode train
It will save the trained model weight at
out/iscnet/a_folder_with_RfD-Net/model_best.pth
Copy the weight path of RfD-Net (see 3. above) into configs/config_files/ISCNet_test.yaml
as
weight: ['out/iscnet/a_folder_with_RfD-Net/model_best.pth']
Run below to output all scenes in the test set.
python main.py --config configs/config_files/ISCNet_test.yaml --mode test
The 3D scenes for visualization are saved in the folder of out/iscnet/a_folder_with_generated_scenes/visualization
. You can visualize a triplet of (input, pred, gt) following a demo below
python utils/scannet/visualization/vis_for_comparison.py
If everything goes smooth, there will be three windows (corresponding to input, pred, gt) popped up by sequence as
Input | Prediction | Ground-truth |
---|---|---|
You can choose each of the following ways for evaluation.
You can export all scenes above to calculate the evaluation metrics with any external library (for researchers who would like to unify the benchmark).
Lower the dump_threshold
in ISCNet_test.yaml
in generation to enable more object proposals for mAP calculation (e.g. dump_threshold=0.05
).
In our evaluation, we voxelize the 3D scenes to keep consistent resolution with the baseline methods. To enable this,
make sure the executable binvox are downloaded and configured as an experiment variable (e.g. export its path in ~/.bashrc for Ubuntu). It will be deployed by Trimesh.
Change the ISCNet_test.yaml
as below for evaluation.
test:
evaluate_mesh_mAP: True
generation:
dump_results: False
Run below to report the evaluation results.
python main.py --config configs/config_files/ISCNet_test.yaml --mode test
The log file will saved in out/iscnet/a_folder_named_with_script_time/log.txt
demo.py
.If you find our work helpful, please consider citing
@InProceedings{Nie_2021_CVPR,
author = {Nie, Yinyu and Hou, Ji and Han, Xiaoguang and Niessner, Matthias},
title = {RfD-Net: Point Scene Understanding by Semantic Instance Reconstruction},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2021},
pages = {4608-4618}
}
RfD-Net is relased under the MIT License. See the LICENSE file for more details.