Our code is tested on the following environment:
- Linux
- Python 3.8
- PyTorch 1.8.1
- Cudatoolkit 11.3
- mmdet3d 0.17.1
PyTorch version 1.8.0 or higher and mmdetection3d==0.17.1.
Setup Environment
conda create -n roadnet python=3.8 -y
conda activate roadnet
conda install pytorch==1.8.1 torchvision==0.9.1 torchaudio==0.8.1 cudatoolkit=11.3 -c pytorch -c conda-forge
Install mmdetection3d correctly. please visit the official documentation. Install MMDetection
pip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/cu111/torch1.8.0/index.html
git clone https://github.com/open-mmlab/mmdetection.git
cd mmdetection
git checkout v2.24.1
sudo pip install -r requirements/build.txt
sudo python3 setup.py develop
cd ..
Install MMSegmentation
pip install mmsegmentation==0.20.2
pip install einops
pip install bezier==0.11.0
Install MMDetection3d
git clone https://github.com/open-mmlab/mmdetection3d.git
cd mmdetection3d
git checkout v0.17.1
sudo pip install -r requirements/build.txt
sudo python3 setup.py develop
cd ..
Add our projects to mmdetection3d projects
cd ${any path outside mmdetection3d}
git clone [email protected]:fudan-zvg/RoadNet.git
cd RoadNet/RoadNetwork-1.8.1/
ln -s {mmdetection3d_path} ./mmdetection3d
mkdir data
ln -s {nuscenes_path} ./data/nuscenes
Please refer to nuScenes for initial preparation
Run the following code to generate .pkl file.
python tools/create_data_centerline.py nuscenes
python tools/create_data_pon_centerline.py nuscenes
mkdir ckpts
Download ResNet-50 Deeplab-V3-Plus checkpoint from MMSegmentation.