install.md

Step-by-step installation instructions

Following https://mmdetection3d.readthedocs.io/en/latest/getting_started.html#installation

a. Create a conda virtual environment and activate it.

conda create -n open-mmlab python=3.8 -y
conda activate open-mmlab

b. Install PyTorch and torchvision following the official instructions.

pip install torch==1.9.1+cu111 torchvision==0.10.1+cu111 torchaudio==0.9.1 -f https://download.pytorch.org/whl/torch_stable.html
# Recommended torch>=1.9

c. Install gcc>=5 in conda env (optional).

conda install -c omgarcia gcc-6 # gcc-6.2

c. Install mmcv-full.

pip install mmcv-full==1.4.0
#  pip install mmcv-full==1.4.0 -f https://download.openmmlab.com/mmcv/dist/cu111/torch1.9.0/index.html

d. Install mmdet and mmseg.

pip install mmdet==2.14.0
pip install mmsegmentation==0.14.1

e. Install mmdet3d from source code.

git clone https://github.com/open-mmlab/mmdetection3d.git
cd mmdetection3d
git checkout v0.17.1 # Other versions may not be compatible.
pip install -v -e .

f. Install timm.

pip install timm

g. Clone VoxFormer.

git clone https://github.com/NVlabs/VoxFormer.git

h. Prepare pretrained resnet50 models.

cd VoxFormer && mkdir ckpts && cd ckpts

Download the pretrained resnet50.

i. Build deformable 3D attention ops.

cd VoxFormer/deform_attn_3d 
python setup.py build_ext --inplace