cuda & MMdetecction3d 安装

MM

安装Anaconda3

安装cuda10.1

wget https://developer.download.nvidia.com/compute/cuda/10.1/Prod/local_installers/cuda_10.1.243_418.87.00_linux.run

sh ****.sh

continue

accept

安装的时候除了cudatoolkit其他都不要安装

(可以百度查一下无root权限自定义路径安装cuda)

重新修改一下bashrc文件并source

然后就可以开始安装了。

安装完毕之后，需要配置一下 .bashrc，在其下方添加：

export PATH=/home//cuda-10.2/bin:$PATH

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/home//cuda-10.2/lib64/

esc + : wq保存退出

保存后运行一下 source .bashrc，就可以用 nvcc -V 试试有没有安装成功了。

安装mm

conda create -n xx python=3.8

conda activate xx

git clone -b v0.18.1 GitHub - open-mmlab/mmdetection3d: OpenMMLab's next-generation platform for general 3D object detection.

pip install torch==1.6.0+cu101 torchvision==0.7.0+cu101 -f <https://download.pytorch.org/whl/torch_stable.html>

pip install mmcv-full==1.3.13 -f https://download.openmmlab.com/mmcv/dist/cu101/torch1.6.0/index.html

pip install mmdet==2.19.0

pip install mmsegmentation==0.20.0

pip install -v -e .

安装完了环境记得把之前有mm的PYTHONPATH都注释掉, 添加上自己新的那个PYTHONPATH, 要不然会找到其他路径里的mmdet3d

环境变量改坏了用这个

SUNRGBD

ln -s /data2/yaoming/fcaf3d/data/sunrgbd/sunrgbd_trainval/ /data2/yaoming/project/mmdetection3d/data/sunrgbd/

python tools/create_data.py sunrgbd --root-path ./data/sunrgbd --out-dir ./data/sunrgbd --extra-tag sunrgbd

ScanNet

ln -s /data2/yaoming/fcaf3d/data/scannet/scannet_instance_data /data2/yaoming/project/mmdetection3d/data/scannet/

ln -s /data2/yaoming/fcaf3d/data/scannet/semantic_mask /data2/yaoming/project/mmdetection3d/data/scannet/

ln -s /data2/yaoming/fcaf3d/data/scannet/seg_info/ /data2/yaoming/project/mmdetection3d/data/scannet

python tools/create_data.py scannet --root-path ./data/scannet --out-dir ./data/scannet --extra-tag scannet