cmake version: 3.22.1
cuda version: 11.5
Abstract
This is a 3D scene reconstruction algorithm that combines monocular dense SLAM and hierarchical voxel neural radiation fields. It can achieve accurate radiation field construction using image sequences in real time and does not require pose or depth input. The core idea of the paper is to use a monocular dense SLAM method to estimate the camera pose and dense depth map and their uncertainty, and use the above information as a supervision signal to train the NeRF scene representation.
1. Update system software source
Backup configuration file /etc/apt/sources.list
sudo cp /etc/apt/sources.list /etc/apt/sources.list.backup //Back up the software source configuration file
Edit configuration file
sudo vim /etc/apt/sources.list //Edit the configuration file and enter the command mode /***Basic usage of vim*** i enters editing (input) mode, at which time you can make changes to the file Esc exits editing mode and enters command mode : Enter the bottom line command mode with a colon before the command. :w save file :q exit edit mode :wq Save file and exit edit mode :q! Quit without saving (forced)
Reference: Ubuntu20.04 replacement software source_ubuntu20.04 software source-CSDN blog
2. Check the cuda version
Install
sudo apt install nvidia-cuda-toolkit
and then use
nvcc -V
You can check the cuda version:
3. Download the NeRF source code (see point 10, this is wrong)
Download the latest NeRF source code first
https://github.com/ToniRV/NeRF-SLAM.git
4. Install Anaconda
Index of /anaconda/archive/ | Tsinghua University Open Source Software Mirror Station | Tsinghua Open Source Mirror
I installed the latest version of https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/Anaconda3-2023.09-0-Linux-x86_64.sh
Install Anaconda3
bash Anaconda3-2022.10-Linux-x86_64.sh
The correct approach is: press Enter first, and then keep pressing s to go to the next page. Then ‘yes’ or ‘no’ will appear, and then press Enter again to install.
Add environment variables
sudo vim ~/.bashrc //Add at the end of the pop-up text //Remember to change username export PATH="/home/username/anaconda3/bin:$PATH"
Everyone’s document is not exactly the same, just add it at the end of the document. Then: wq save and exit, source it to take effect.
source ~/.bashrc //Check whether the installation is successful or not conda --version
5. Add Tsinghua mirror source
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/r conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/pro conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/msys2 conda config --set show_channel_urls yes
6. Install dependencies
Create and activate a virtual environment
conda create -n NeRF_SLAM source activate NeRF_SLAM
7. Install cmake3.22.6
Cmake version is required to be above 3.22, please refer to this blog
Install cmake3.22.6, https://cmake.org/files/v3.22/cmake-3.22.6-linux-x86_64.tar.gz, domestic source: Index of /files (cmake.org)
Delete old version
sudo apt autoremove cmake
The following errors may occur
error8.1 bash: /usr/bin/cmake: is a directory
error8.2 bash: /usr/bin/cmake: No such file or directory
error8.3 bash: /usr/bin/cmake: too many levels of symbolic links
error8.4 ln: cannot create symbolic link ‘/usr/bin/cmake’: file already exists
Solution:
8. Install pytorch
The version corresponding to cuda can be viewed with nvcc -V
Note that pytorch needs to download the version corresponding to cuda.
//CUDA 11.7 conda install pytorch torchvision torchaudio pytorch-cuda=11.7 -c pytorch -c nvidia //Pip install requirements: pip install -r requirements.txt
During the installation process, this will appear:
The following packages will be SUPERSEDED by a higher-priority channel: ca-certificates pkgs/main --> anaconda/pkgs/main certifi pkgs/main --> anaconda/pkgs/main conda pkgs/main --> anaconda/pkgs/main Proceed ([y]/n)?
My choice is yes, refer to What does The following packages will be SUPERCEDED by a higher-priority channel mean? – Missing Qianxun – Blog Garden (cnblogs.com)
Just replaces the priority of the source
9. Start compiling NerRF_SLAM
In the NeRF_SLAM directory, run
pip install -r requirements.txt //Install required dependencies pip install -r ./thirdparty/gtsam/python/requirements.txt //Install required dependencies pip install -r ./thirdparty/instant-ngp/requirements.txt //Install required dependencies
When NeRF_SLAM runs and installs dependencies, a problem occurs:
error: subprocess-exited-with-error × python setup.py egg_info did not run successfully. │ exit code: 1 ╰─> [6 lines of output] Traceback (most recent call last): File “
Solution:
Check whether torch is installed correctly: First make sure that PyTorch has been installed in your environment, and that the version of PyTorch is compatible with torch-scatter. You can install PyTorch using pip or conda, depending on your environment and needs.
For example, use conda to install PyTorch:
conda install pytorch torchvision torchaudio -c pytorch
Enter the NeRF_SLAM directory and run
error occurs
ERROR: Ignored the following versions that require a different python version: 1.21.2 Requires-Python >=3.7,<3.11; 1.21.3 Requires-Python >=3.7,<3.11; 1.21.4 Requires-Python >=3.7,< 3.11; 1.21.5 Requires-Python >=3.7,<3.11; 1.21.6 Requires-Python >=3.7,<3.11
ERROR: Could not find a version that satisfies the requirement numpy~=1.21.2 (from versions: 1.3.0, 1.4.1, 1.5.0, 1.5.1, 1.6.0, 1.6.1, 1.6.2, 1.7 .0, 1.7.1, 1.7.2, 1.8.0, 1.8.1, 1.8.2, 1.9.0, 1.9.1, 1.9.2, 1.9.3, 1.10.0.post2, 1.10.1, 1.10 .2, 1.10.4, 1.11.0, 1.11.1, 1.11.2, 1.11.3, 1.12.0, 1.12.1, 1.13.0, 1.13.1, 1.13.3, 1.14.0, 1.14.1 , 1.14.2, 1.14.3, 1.14.4, 1.14.5, 1.14.6, 1.15.0, 1.15.1, 1.15.2, 1.15.3, 1.15.4, 1.16.0, 1.16.1, 1.16 .2, 1.16.3, 1.16.4, 1.16.5, 1.16.6, 1.17.0, 1.17.1, 1.17.2, 1.17.3, 1.17.4, 1.17.5, 1.18.0, 1.18.1 , 1.18.2, 1.18.3, 1.18.4, 1.18.5, 1.19.0, 1.19.1, 1.19.2, 1.19.3, 1.19.4, 1.19.5, 1.20.0, 1.20.1, 1.20 .2, 1.20.3, 1.21.0, 1.21.1, 1.22.0, 1.22.1, 1.22.2, 1.22.3, 1.22.4, 1.23.0rc1, 1.23.0rc2, 1.23.0rc3, 1.23.0 , 1.23.1, 1.23.2, 1.23.3, 1.23.4, 1.23.5, 1.24.0rc1, 1.24.0rc2, 1.24.0, 1.24.1, 1.24.2, 1.24.3, 1.24.4, 1.25 .0rc1, 1.25.0, 1.25.1, 1.25.2, 1.26.0b1, 1.26.0rc1, 1.26.0, 1.26.1)
ERROR: No matching distribution found for numpy~=1.21.2
(wrong) solution:
conda install numpy=1.21.2
If you do this, you will get an error later:
ERROR: pip’s dependency resolver does not currently take into account all the packages that are installed. This behavior is the source of the following dependency conflicts.
pandas 2.1.1 requires numpy>=1.22.4; python_version < "3.11", but you have numpy 1.21.6 which is incompatible.
(Correct) solution:
conda install numpy=1.22.4
Compile instant-ngp (this step is the most difficult)
Play it by ear! ! ! ! ! ! ! I put all the ones I used below for reference.
Entering the next step, an error occurred when installing instant-ngp.
cmake ./thirdparty/instant-ngp -B build_ngp
I keep getting errors:
CMake Error at CMakeLists.txt:42 (message):
Some instant-ngp dependencies are missing. If you forgot the “–recursive”
flag when cloning this project, this can be fixed by calling “git submodule
update –init –recursive”.
Reference 1: Ubuntu installation Ceres-Solver_Check whether ceres is installed according to z_Li Changhao’s blog-CSDN blog
Reference 2: Installation – Ceres Solver (ceres-solver.org)
Reference 3: ubuntu20.04 nerf Instant-ngp-CSDN Blog
Reference 4: If you cannot install cere2.2.0, install cere2.0.0 http://ceres-solver.org/ceres-solver-2.0.0.tar.gz
Reference 5: Ubuntu22.04 installs ceres-solver, cmake compilation error reports tbb_stddef.h-CSDN Blog
I was stupid. I didn’t know how to fix it in the end. The instant-ngp I finally used was: NVlabs/instant-ngp: Instant neural graphics primitives: lightning fast NeRF and more (github.com)
So annoying! ! ! ! ! It troubled me for two or three days! ! ! ! !
cmake --build build_ngp --config RelWithDebInfo -j2
OK! Or not
Compile gtsam
First download https://github.com/ToniRV/gtsam-1.git
Then replace gtsam in the NeRF_SLAM/thirdparty directory and change the name to gtsam
cmake ./thirdparty/gtsam -DGTSAM_BUILD_PYTHON=1 -B build_gtsam cmake --build build_gtsam --config RelWithDebInfo -j cd build_gtsam make python-install cdpython python setup.py install
This one is not difficult, but it takes a long time
OK! I start from scratch! Because I originally downloaded NeRF_SLAM from windows and ssh it to linux, I found that doing this might cause some problems! It’s just that some dependencies were not downloaded! So this time!
10. Re-download the source code
git clone https://github.com/ToniRV/NeRF-SLAM.git --recurse-submodules git submodule update --init --recursive
I think! Success is in sight! ! ! ! ! !
What a difference! Neither instant-ngp nor gtsam in the downloaded thirdparty are blank folders! !
In this way, you only need to replace gtsam.
Destroy, go*