a. Create a conda virtual environment and activate it.
conda create -n occ python=3.7 -y
conda activate occ
1. Install PyTorch and torchvision following the official instructions.
Pytorch and cuda with higher version is also supported.
conda install pytorch==1.10.0 torchvision==0.11.0 torchaudio==0.10.0 cudatoolkit=10.2 -c pytorch
2. Install mmcv-full.
pip install mmcv-full==1.4.1
3. Install mmdet and mmseg.
pip install mmdet==2.19.0
pip install mmsegmentation==0.20.0
4. Install mmdet3d from source code.
git clone https://github.com/open-mmlab/mmdetection3d.git
cd mmdetection3d
git checkout v0.18.1 # Other versions may not be compatible.
python setup.py install
5. Install timm.
pip install timm
6. Clone OccNet.
git clone https://github.com/OpenDriveLab/OccNet.git
7. Prepare pretrained model
cd OccNet
mkdir ckpts
cd ckpts
wget https://github.com/zhiqi-li/storage/releases/download/v1.0/r101_dcn_fcos3d_pretrain.pth
8. Install InternImage (Optional).
OccNet support InternImage backbone with much better performance than ImageNet backbone.
- Install dcnv3
cd OccNet
cd projects/mmdet3d_plugin/bevformer/backbones/ops_dcnv3
sh make.sh
python test.py
- Prepare pretrained model
cd OccNet/ckpts
wget https://huggingface.co/OpenGVLab/InternImage/resolve/main/mask_rcnn_internimage_s_fpn_3x_coco.pth