mmdet3d

mmdetection3d安装

export PATH="/home/jsz/anaconda3/bin:$PATH"
source ~/.bashrc

conda create -n qswl python=3.7
source activate qswl
pip install openmim
mim install mmdet3d

a. Create a conda virtual environment and activate it.

conda create -n open-mmlab python=3.7 -y
source activate open-mmlab

b. Install PyTorch and torchvision following the official instructions.

conda install pytorch==1.5.0 cudatoolkit=10.1 torchvision==0.6.0 -c pytorch

c. Install MMCV. mmcv-full is necessary since MMDetection3D relies on MMDetection, CUDA ops in mmcv-full are required.

预编译好的

pip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/{cu_version}/{torch_version}/index.html

例如 CUDA 10.1 和 PyTorch 1.5.0环境

pip install mmcv-full -f https://download.openmmlab.com/mmcv/dist/cu101/torch1.5.0/index.html

d. Install MMDetection.

pip install mmdet==2.14.0

e. Install MMSegmentation.

pip install mmsegmentation==0.14.1

f. Clone the MMDetection3D repository.

git clone https://github.com/open-mmlab/mmdetection3d.git
cd mmdetection3d

rz mm.zip
unzip mm.zip

g.Install build requirements and then install MMDetection3D.

pip install -v -e .  # or "python setup.py develop"

记录：最后这个没搜索到：安装完成后提示：Removed build tracker: ‘/tmp/pip-req-tracker-7d3rtjzs’

demo

python demo/pcd_demo.py demo/data/kitti/kitti_000008.bin configs/second/hv_second_secfpn_6x8_80e_kitti-3d-car.py ~/Downloads/mmdetection3d/checkpoints/hv_second_secfpn_6x8_80e_kitti-3d-car_20200620_230238-393f000c.pth

训练