0
0

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?

@xtrizeShino(Shinobu HUYUGIRI)

Ubuntu22.04に主要な機械学習のフレームワークをインストールする(2026年対応版)

0
Last updated at Posted at 2026-09-05

以下、真っ白なUbuntu22.04 LTSに
2026/09/05で競合や依存性不良を発生させず
機械学習用のフレームワーク一式をインストールする手順です。
本手順により以下が有効になります

  • NVIDIAドライバ (595-open)
  • CUDA12.6 ToolKit (依存関係により13.3は断念)
  • cuDNN9.24.1
  • Docker + NVIDIA Container Toolkit
  • NGC CLI
  • GPUを利用可能なPython仮想環境
    • TensorFlow + ONNX
    • TensorRT + ONNX
    • PyTorch + ONNX
  • NVIDIA TAO / Samples
  • DeepStream

以上です。

Ubuntu22.04のインストール

https___qiita-user-contents.imgix.net_https2F2F02F0b35301c-d7fd-4b4f-fdd8-d955714ce393.jpeg_ixlib=rb-4.0.avif

機械学習用のフレームワーク一式をインストールする

リモート接続環境とエディタをインストールする

### ホームディレクトリへ移動
$ cd ~

### ホームディレクトリ名から日本語を除外する
$ LANG=C xdg-user-dirs-gtk-update

### ユーザーからsudoをパスワード無しで実行できるようにする
$ sudo vi /etc/sudoers
# -->> 以下の行を追加する(ユーザー名がshinoの場合)
# shino   ALL=(ALL:ALL) NOPASSWD:ALL

### 再起動する
$ sudo reboot

# リポジトリにアクセス
$ sudo apt update

# 更新可能なパッケージをすべてアップグレード
$ sudo apt upgrade

# 再起動
$ sudo reboot

### sshサーバをインストールします
$ sudo apt install openssh-server

### Ubuntu24.04からsshdでなくsshになっています 
# ... 起動時にサービスが開始されるように設定します
$ sudo systemctl start ssh
$ sudo systemctl enable ssh

### vimエディタをインストールします。
$ sudo apt install vim

### 鍵の登録
$ mkdir -p /home/shino/.ssh
$ touch /home/shino/.ssh/authorized_keys
$ sudo chmod 700 /home/shino/.ssh
$ sudo chmod 640 /home/shino/.ssh/authorized_keys

### 公開鍵を登録する
$ vi /home/shino/.ssh/authorized_keys
# -->> Windows上の "id_rsa.pub" の内容を書き込む

### 再起動する
$ sudo reboot

グラフィックドライバをインストールする

# Linuxカーネルのヘッダをインストールする
$ sudo apt update
$ sudo apt install linux-headers-$(uname -r)

### インストール可能なGPUドライバのリストを表示する
$ ubuntu-drivers devices
#== /sys/devices/pci0000:00/0000:00:1c.4/0000:05:00.0 ==
#modalias : pci:v000010DEd00002489sv0000103Csd000088D4bc03sc00i00
#vendor   : NVIDIA Corporation
#model    : GA104 [GeForce RTX 3060 Ti Lite Hash Rate]
#driver   : nvidia-driver-610-open - distro non-free
#driver   : nvidia-driver-580-open - distro non-free
#driver   : nvidia-driver-610 - distro non-free
#driver   : nvidia-driver-580 - distro non-free
#driver   : nvidia-driver-595-open - distro non-free recommended
#driver   : nvidia-driver-580-server - distro non-free
#driver   : nvidia-driver-545-open - distro non-free
#driver   : nvidia-driver-580-server-open - distro non-free
#driver   : nvidia-driver-545 - distro non-free
#driver   : nvidia-driver-470-server - distro non-free
#driver   : nvidia-driver-595-server - distro non-free
#driver   : nvidia-driver-595-server-open - distro non-free
#driver   : nvidia-driver-595 - distro non-free
#driver   : xserver-xorg-video-nouveau - distro free builtin

### recommendedをインストールする
$ sudo apt install nvidia-driver-595-open
$ sudo update-initramfs -u
$ sudo reboot

### NVIDIAドライバが適用されていて、すべてのGPUが見えるか確認
$ nvidia-smi
#+-----------------------------------------------------------------------------------------+
#| NVIDIA-SMI 595.84                 Driver Version: 595.84         CUDA Version: 13.2     |
#+-----------------------------------------+------------------------+----------------------+
#| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
#| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
#|                                         |                        |               MIG M. |
#|=========================================+========================+======================|
#|   0  NVIDIA GeForce RTX 3060 Ti     Off |   00000000:01:00.0  On |                  N/A |
#| 31%   40C    P8              8W /  200W |     223MiB /   8192MiB |      2%      Default |
#|                                         |                        |                  N/A |
#+-----------------------------------------+------------------------+----------------------+
#|   1  NVIDIA GeForce RTX 3060 Ti     Off |   00000000:05:00.0 Off |                  N/A |
#|  0%   48C    P8              8W /  200W |      15MiB /   8192MiB |      0%      Default |
#|                                         |                        |                  N/A |
#+-----------------------------------------+------------------------+----------------------+
#
#+-----------------------------------------------------------------------------------------+
#| Processes:                                                                              |
#|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
#|        ID   ID                                                               Usage      |
#|=========================================================================================|
#|    0   N/A  N/A            1476      G   /usr/lib/xorg/Xorg                      106MiB |
#|    0   N/A  N/A            1767      G   /usr/bin/gnome-shell                     80MiB |
#|    1   N/A  N/A            1476      G   /usr/lib/xorg/Xorg                        4MiB |
#+-----------------------------------------------------------------------------------------+

CUDA12.6とcuDNNをインストールする

### CUDA 12.6 をインストールする
# TensorFlowが13.xを完全サポートしていないため
# DeepStreamがCUDA12.8をサポートしていないため
# https://developer.nvidia.com/cuda-12-6-3-download-archive?target_os=Linux&target_arch=x86_64&Distribution=Ubuntu&target_version=22.04&target_type=deb_network

$ wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-keyring_1.1-1_all.deb
$ sudo dpkg -i cuda-keyring_1.1-1_all.deb
$ sudo apt-get update
$ sudo apt-get -y install cuda-toolkit-12-6
$ sudo apt-mark hold cuda-toolkit-12-6
###
# 注意：Driver Installerの手順は実行しない
# 595-openから変わってしまうため
###

# 最後の行に下記の4行を追加
$ vi ~/.bashrc
export PATH="/usr/local/cuda/bin:$PATH"
export PATH="~/.local/bin:$PATH"
export LD_LIBRARY_PATH="/usr/local/cuda/lib64:$LD_LIBRARY_PATH"
export CUDA_PATH="/usr/local/cuda-12.6"

# 設定した環境変数を反映
$ source ~/.bashrc

# Tool-Kitのコマンドにパスが通っているか確認
$ nvcc -V
#nvcc: NVIDIA (R) Cuda compiler driver
#Copyright (c) 2005-2024 NVIDIA Corporation
#Built on Tue_Oct_29_23:50:19_PDT_2024
#Cuda compilation tools, release 12.6, V12.6.85
#Build cuda_12.6.r12.6/compiler.35059454_0

### cuDNN 9.24.1 をインストールする
# 2026/09/04最新
# https://developer.nvidia.com/cudnn-downloads?target_os=Linux&target_arch=x86_64&Distribution=Ubuntu&target_version=22.04&target_type=deb_network&Configuration=Full
## $ wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2204/x86_64/cuda-keyring_1.1-1_all.deb
## $ sudo dpkg -i cuda-keyring_1.1-1_all.deb
$ sudo apt-get update

# CUDA12.x決め打ちでインストールする
$ sudo apt-get -y install cudnn9-cuda-12
$ sudo apt-mark hold cudnn9-cuda-12

# このあたりで一旦再起動
$ sudo reboot

Dockerをインストールする

### Dockerをインストールする
# https://docs.docker.com/engine/install/ubuntu/

# まず削除
$ sudo apt remove $(dpkg --get-selections docker.io docker-compose docker-compose-v2 docker-doc docker-buildx podman-docker containerd runc | cut -f1)

# インストールの準備
$ sudo apt update
$ sudo apt install ca-certificates curl
$ sudo install -m 0755 -d /etc/apt/keyrings
$ sudo curl -fsSL https://download.docker.com/linux/ubuntu/gpg -o /etc/apt/keyrings/docker.asc
$ sudo chmod a+r /etc/apt/keyrings/docker.asc

# Add the repository to Apt sources:
$ sudo tee /etc/apt/sources.list.d/docker.sources <<EOF
Types: deb
URIs: https://download.docker.com/linux/ubuntu
Suites: $(. /etc/os-release && echo "${UBUNTU_CODENAME:-$VERSION_CODENAME}")
Components: stable
Architectures: $(dpkg --print-architecture)
Signed-By: /etc/apt/keyrings/docker.asc
EOF

$ sudo apt update

# 最新版をインストール
$ sudo apt install docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin

# 非rootユーザーもDockerを操作できるようにする
$ sudo groupadd docker
$ sudo usermod -aG docker $USER
$ newgrp docker

# 再起動
$ sudo reboot

# Dockerのテスト
$ docker run --rm hello-world


### NVIDIA Container Toolkitをインストールする
# Dockerのコンテナ内からGPUにアクセスする
# https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/latest/install-guide.html
$ sudo apt update && sudo apt install -y --no-install-recommends \
   ca-certificates \
   curl \
   gnupg2

$ curl -fsSL https://nvidia.github.io/libnvidia-container/gpgkey | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-container-toolkit-keyring.gpg \
  && curl -s -L https://nvidia.github.io/libnvidia-container/stable/deb/nvidia-container-toolkit.list | \
    sed 's#deb https://#deb [signed-by=/usr/share/keyrings/nvidia-container-toolkit-keyring.gpg] https://#g' | \
    sudo tee /etc/apt/sources.list.d/nvidia-container-toolkit.list

# インストール
$ sudo apt update
$ export NVIDIA_CONTAINER_TOOLKIT_VERSION=1.20.0-1
  sudo apt install -y \
      nvidia-container-toolkit=${NVIDIA_CONTAINER_TOOLKIT_VERSION} \
      nvidia-container-toolkit-base=${NVIDIA_CONTAINER_TOOLKIT_VERSION} \
      libnvidia-container-tools=${NVIDIA_CONTAINER_TOOLKIT_VERSION} \
      libnvidia-container1=${NVIDIA_CONTAINER_TOOLKIT_VERSION}

# コンテナのランタイムがDockerであることを登録する
$ sudo nvidia-ctk runtime configure --runtime=docker
# INFO[0000] Config file does not exist; using empty config
# INFO[0000] Wrote updated config to /etc/docker/daemon.json
# INFO[0000] It is recommended that docker daemon be restarted.

# Dockerサービスを再起動する
$ sudo systemctl restart docker

# GPUが見えなくなる問題の対策
# -->> https://qiita.com/tttamaki/items/4f369633b7f53f5131a1
$ sudo vi /etc/docker/daemon.json
{   
    "runtimes": { 
        "nvidia": { 
            "args": [],
            "path": "nvidia-container-runtime"
        }
    },
    "exec-opts": ["native.cgroupdriver=cgroupfs"]
}

# Dockerサービスを再起動する
$ sudo systemctl restart docker

# Dockerにプロセス管理機能cgroup設定が結びつけられていることを確認する
$ docker info | grep Cgroup
# Cgroup Driver: cgroupfs
# Cgroup Version: 2

# Dockerコンテナを取得して、コンテナにすべてのGPUを接続する
$ docker run --rm -it --gpus all pytorch/pytorch:2.6.0-cuda12.6-cudnn9-devel

(on Conatiner) $ /usr/local/cuda/bin/nvcc --version
#nvcc: NVIDIA (R) Cuda compiler driver
#Copyright (c) 2005-2024 NVIDIA Corporation
#Built on Tue_Oct_29_23:50:19_PDT_2024
#Cuda compilation tools, release 12.6, V12.6.85
#Build cuda_12.6.r12.6/compiler.35059454_0

(on Conatiner) $ nvidia-smi
#Fri Sep  4 13:54:48 2026
#+-----------------------------------------------------------------------------------------+
#| NVIDIA-SMI 595.84                 Driver Version: 595.84         CUDA Version: 13.2     |
#+-----------------------------------------+------------------------+----------------------+
#| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
#| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
#|                                         |                        |               MIG M. |
#|=========================================+========================+======================|
#|   0  NVIDIA GeForce RTX 3060 Ti     Off |   00000000:01:00.0  On |                  N/A |
#| 31%   32C    P8              8W /  200W |     222MiB /   8192MiB |      0%      Default |
#|                                         |                        |                  N/A |
#+-----------------------------------------+------------------------+----------------------+
#|   1  NVIDIA GeForce RTX 3060 Ti     Off |   00000000:05:00.0 Off |                  N/A |
#|  0%   35C    P8              7W /  200W |      15MiB /   8192MiB |      0%      Default |
#|                                         |                        |                  N/A |
#+-----------------------------------------+------------------------+----------------------+
#
#+-----------------------------------------------------------------------------------------+
#| Processes:                                                                              |
#|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
#|        ID   ID                                                               Usage      |
#|=========================================================================================|
#|  No running processes found                                                             |
#+-----------------------------------------------------------------------------------------+

(on Conatiner) $ exit

# 確認完了

NVIDIAのリソースを取得するNGCをインストールする

### NGC CLIのセットアップ
# https://org.ngc.nvidia.com/setup/installers/cli
$ cd ~
# ツールのダウンロード
$ wget --content-disposition https://api.ngc.nvidia.com/v2/resources/nvidia/ngc-apps/ngc_cli/versions/3.59.0/files/ngccli_linux.zip -O ngccli_linux.zip && unzip ngccli_linux.zip
# 正常チェック
$ find ngc-cli/ -type f -exec md5sum {} + | LC_ALL=C sort | md5sum -c ngc-cli.md5
# NGC CLIに実行権限を付与してパスを通す
$ chmod u+x ngc-cli/ngc
$ echo "export PATH=\"\$PATH:$(pwd)/ngc-cli\"" >> ~/.bashrc && source ~/.bashrc

###
# https://org.ngc.nvidia.com/setup
###
$ ngc config set
#CLI_VERSION: Latest - 4.34.10 available (current: 3.59.0). Please update by using the command 'ngc version upgrade'
#
#Enter API key [no-apikey]. Choices: [<VALID_APIKEY>, 'no-apikey']: nvapi-【ここは自身のアカウントより取得する】
#Enter CLI output format type [ascii]. Choices: ['ascii', 'csv', 'json']:
#Enter org [no-org]. Choices: ['【ここは自身のアカウントが表示される】']: 【ここは自身のアカウントを指定する】
#Enter team [no-team]. Choices: ['no-team']:
#Enter ace [no-ace]. Choices: ['no-ace']:
#Validating configuration...
#Successfully validated configuration.
#Saving configuration...
#Successfully saved NGC configuration to /home/shino/.ngc/config

# Dockerコンテナをダウンロードする際に利用するNGC CLIの設定
$ docker login nvcr.io
#Username: $oauthtoken
#Password: nvapi-【ここは自身のアカウントより取得する】
#
#WARNING! Your credentials are stored unencrypted in '/home/shino/.docker/config.json'.
#Configure a credential helper to remove this warning. See
#https://docs.docker.com/go/credential-store/
#
#Login Succeeded

開発用のPython仮想環境を準備する

TensorRTをインストールする

## $ wget https://developer.nvidia.com/downloads/compute/machine-learning/tensorrt/10.6.0/local_repo/nv-tensorrt-local-repo-ubuntu2204-10.6.0-cuda-12.6_1.0-1_amd64.deb
## $ sudo dpkg -i nv-tensorrt-local-repo-ubuntu2204-10.6.0-cuda-12.6_1.0-1_amd64.deb
## $ sudo cp /var/nv-tensorrt-local-repo-ubuntu2204-10.6.0-cuda-12.6/*-keyring.gpg /usr/share/keyrings/
$ sudo apt update
$ trt_version="10.6.0.26-1+cuda12.6"
$ sudo apt-get install -y \
  libnvinfer10="${trt_version}" \
  libnvinfer-plugin10="${trt_version}" \
  libnvonnxparsers10="${trt_version}"

$ sudo bash -c 'cat > /etc/apt/preferences.d/cuda-12-6.pref <<EOF
Package: *
Pin: origin developer.download.nvidia.com
Pin-Priority: 1001
EOF'
$ sudo apt update
$ sudo apt install -y libnvinfer-vc-plugin10=${trt_version}
$ sudo apt-get install -y python3-libnvinfer="${trt_version}"

$ sudo apt-mark hold \
  libnvinfer10 \
  libnvinfer-plugin10 \
  libnvonnxparsers10 \
  libnvinfer-vc-plugin10 \
  python3-libnvinfer

# このへんで再起動
$ sudo reboot

Python仮想環境をインストールする

### Pythonの仮想環境モジュール"venv"をインストール
$ sudo apt update
$ sudo apt install python3.10-venv python3-pip

TensorRTの実行環境を準備する

$ mkdir ~/env
$ python3 -m venv ~/env/trt --system-site-packages
$ source ~/env/trt/bin/activate
(trt) $ pip install --upgrade pip setuptools wheel
(trt) $ pip install \
  pandas scipy matplotlib seaborn scikit-learn opencv-python \
  "numpy>=1.24,<2" \
  "cuda-python==12.6" \
  "onnx==1.16.1" \
  "onnxruntime-gpu>=1.21,<1.27"

### pythonからTensorRTを呼び出してみる
(trt) $ python --version
# Python 3.10.12

(trt) $ python -c "
import tensorrt as trt
print(trt.__version__)
assert trt.Builder(trt.Logger())
"
#10.6.0

(trt) $ python -c "
import onnxruntime as ort
print('ONNX Runtime version:', ort.__version__)
print('Available providers:', ort.get_available_providers())
"
#ONNX Runtime version: 1.23.2
#Available providers: ['TensorrtExecutionProvider', 'CUDAExecutionProvider', 'CPUExecutionProvider']
# -->> CUDAが見えていれば成功！

(trt) $ deactivate
$

TensorFlowの実行環境を準備する

######
### TensorFlow-gpuの仮想環境を作成する
######
$ python3 -m venv ~/env/tf_onnx
$ source ~/env/tf_onnx/bin/activate
$ pip install --upgrade pip setuptools wheel
$ pip install \
  "tensorflow[and-cuda]==2.15.1" \
  "tf2onnx==1.17.0" \
  "onnx==1.16.1" \
  "onnxruntime-gpu>=1.21,<1.27" \
  "numpy==1.26.4" \
  pandas scipy scikit-learn matplotlib seaborn opencv-python

(tf_onnx) $ python -c "
import tensorflow as tf
print('TensorFlow version:', tf.__version__)
print('Built with CUDA:', tf.test.is_built_with_cuda())
print('GPU devices:', tf.config.list_physical_devices('GPU'))
"
#2026-09-12 00:04:43.376493: I tensorflow/core/util/port.cc:113] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`.
#2026-09-12 00:04:43.397286: E external/local_xla/xla/stream_executor/cuda/cuda_dnn.cc:9261] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
#2026-09-12 00:04:43.397311: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:607] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
#2026-09-12 00:04:43.397843: E external/local_xla/xla/stream_executor/cuda/cuda_blas.cc:1515] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
#2026-09-12 00:04:43.401076: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
#To enable the following instructions: AVX2 AVX_VNNI FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
#2026-09-12 00:04:43.743121: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:38] TF-TRT Warning: Could not find TensorRT
#TensorFlow version: 2.15.1
#Built with CUDA: True
#2026-09-12 00:04:44.329172: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
#2026-09-12 00:04:44.331557: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
#2026-09-12 00:04:44.452212: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
#2026-09-12 00:04:44.454311: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
#2026-09-12 00:04:44.456418: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
#2026-09-12 00:04:44.458419: I external/local_xla/xla/stream_executor/cuda/cuda_executor.cc:901] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
#GPU devices: [PhysicalDevice(name='/physical_device:GPU:0', device_type='GPU'), PhysicalDevice(name='/physical_device:GPU:1', device_type='GPU')]
# -->> GPUリソースが見えていれば成功！

(tf_onnx) $ python -c "
import onnxruntime as ort
print('ONNX Runtime version:', ort.__version__)
print('Available providers:', ort.get_available_providers())
"
#ONNX Runtime version: 1.23.2
#Available providers: ['TensorrtExecutionProvider', 'CUDAExecutionProvider', 'CPUExecutionProvider']
# -->> CUDAが見えていれば成功！

(tf_onnx) $ deactivate
$

PyTorchの実行環境を準備する

###### 
### PyTorchの実行環境を構築してGPUが使えるか確認
###### 
$ python3 -m venv ~/env/torch
$ source ~/env/torch/bin/activate

(torch) $ python -m pip install --upgrade pip setuptools wheel

# PyTorch (CUDA 12.6)
(torch) $ pip install \
  "torch==2.11.0" \
  "torchvision==0.26.0" \
  "torchaudio==2.11.0" \
  --index-url https://download.pytorch.org/whl/cu126

# ONNX/ORT
(torch) $ pip install \
  "onnx==1.22.0" \
  "onnxscript==0.7.2" \
  "onnxruntime-gpu==1.23.2"

# Utility packages
(torch) $ pip install \
  "numpy==2.2.6" \
  pandas \
  scipy \
  scikit-learn \
  matplotlib \
  seaborn \
  opencv-python \
  six \
  torchinfo

(torch) $ pip check
#No broken requirements found.
  
# GPUが見えるか確認
(torch) $ python -c "
import torch
print('PyTorch version:', torch.__version__)
print('CUDA available:', torch.cuda.is_available())
print('CUDA version (PyTorch):', torch.version.cuda)
if torch.cuda.is_available():
    print('GPU name:', torch.cuda.get_device_name(0))
    print('GPU count:', torch.cuda.device_count())
    x = torch.rand(5, 3).cuda()
    print('Tensor on GPU:', x)
"
#CUDA available: True
#CUDA version (PyTorch): 12.6
#GPU name: NVIDIA GeForce RTX 3060 Ti
#GPU count: 2
#Tensor on GPU: tensor([[0.4332, 0.2673, 0.2932],
#        [0.7222, 0.8852, 0.0042],
#        [0.9091, 0.9655, 0.2439],
#        [0.4586, 0.4896, 0.3815],
#        [0.8671, 0.5108, 0.9195]], device='cuda:0')

(torch) $ python -c "
import onnxruntime as ort
print('ONNX Runtime version:', ort.__version__)
print('Available providers:', ort.get_available_providers())
"
#ONNX Runtime version: 1.23.2
#Available providers: ['TensorrtExecutionProvider', 'CUDAExecutionProvider', 'CPUExecutionProvider']
# -->> CUDAが見えていれば成功！
(torch) $ deactivate
$

# このへんで再起動
$ sudo reboot

PyTorchのモデルをTFLiteへ書き出すためのコンテナを準備する

### ONNX2TFの準備をする

$ docker pull pinto0309/onnx2tf:latest

#$ docker run --rm --user $(id -u):$(id -g) \
#  -v $(pwd):/work -w /work \
#  pinto0309/onnx2tf:latest \
#  onnx2tf -i model.onnx \
#          -o output_uint8 \
#          -oiqt \
#          -iqd int8 \
#          -oqd int8 \
#          -tb tf_converter \
#          -cind "x" "calib_data.npy" "[[[[0.0,0.0,0.0]]]]" "[[[[1.0,1.0,1.0]]]]"

NVIDIA TAOをインストールする

### TAO Launcher のインストール
$ mkdir -p ~/tao
$ cd ~/tao
$ python3 -m venv tao_env
$ source tao_env/bin/activate
(tao_env) $ pip install --upgrade pip setuptools wheel
(tao_env) $ pip install nvidia-tao
(tao_env) $ tao --help
#~/.tao_mounts.json wasn't found. Falling back to obtain mount points and docker configs from ~/.tao_mounts.json.
#Please note that this will be deprecated going forward.
#usage: tao [-h] {list,stop,info,dataset,deploy,model} ...
#
#Launcher for TAO Toolkit.
#
#options:
#  -h, --help            show this help message and exit
#
#task_groups:
#  {list,stop,info,dataset,deploy,model}

# notebookのインストール
(tao_env) $ pip install jupyter notebook jupyterlab ipykernel
(tao_env) $ python -m ipykernel install --user --name tao_env --display-name "tao_env"
#Installed kernelspec tao_env in /home/shino/.local/share/jupyter/kernels/tao_env

(tao_env) $ cd ~/tao
(tao_env) $ git clone https://github.com/NVIDIA/tao_tutorials.git
(tao_env) $ cd tao_tutorials
(tao_env) $ ls ./setup
#quickstart_launcher.sh  tao-docker-compose

(tao_env) $ bash setup/quickstart_launcher.sh --install
#...
#By installing the TAO Toolkit CLI, you accept the terms and conditions of this license: https://developer.nvidia.com/tao-toolkit-software-license-agreement
#Would you like to continue? (y/n): y
#...
#format_version: 3
#toolkit_version: 6.26.3
#published_date: 03/20/2026
(tao_env) $ cd ~/tao/tao_tutorials/
(tao_env) $ jupyter notebook --ip 0.0.0.0 --port 8888 --allow-root

# このへんで再起動
(tao_env) $ sudo reboot

DeepStreamをインストールする

# CUDA
$ nvcc --version
#nvcc: NVIDIA (R) Cuda compiler driver
#Copyright (c) 2005-2024 NVIDIA Corporation
#Built on Tue_Oct_29_23:50:19_PDT_2024
#Cuda compilation tools, release 12.6, V12.6.85
#Build cuda_12.6.r12.6/compiler.35059454_0

# TensorRT
$ dpkg -l | grep -E 'nvinfer|tensorrt'
#hi  libnvinfer-plugin10                        10.6.0.26-1+cuda12.6                             amd64        TensorRT plugin libraries
#hi  libnvinfer-vc-plugin10                     10.6.0.26-1+cuda12.6                             amd64        TensorRT vc-plugin library
#hi  libnvinfer10                               10.6.0.26-1+cuda12.6                             amd64        TensorRT runtime libraries
#hi  python3-libnvinfer                         10.6.0.26-1+cuda12.6                             amd64        Python 3 bindings for TensorRT standard runtime

# GPUドライバ
$ dpkg -l | grep -E 'nvidia-driver'
#ii  nvidia-driver-595-open                     595.91.07-0ubuntu0.22.04.1                       amd64        NVIDIA driver (open kernel) metapackage

####
#Ubuntu 22.04
#CUDA 12.6
#TensorRT 10.3.0.26 以降
#NVIDIA Driver 560以降
#を確認する
####

# 必要なパッケージをインストールする
$ sudo apt update
$ sudo apt install -y \
  libssl3 \
  libssl-dev \
  libgles2-mesa-dev \
  libgstreamer1.0-0 \
  gstreamer1.0-tools \
  gstreamer1.0-plugins-good \
  gstreamer1.0-plugins-bad \
  gstreamer1.0-plugins-ugly \
  gstreamer1.0-libav \
  libgstreamer-plugins-base1.0-dev \
  libgstrtspserver-1.0-0 \
  libjansson4 \
  libyaml-cpp-dev \
  libjsoncpp-dev \
  protobuf-compiler \
  gcc \
  make \
  git \
  python3

# Ububtu22.04のglibcを更新する
$ sudo apt install meson ninja-build
$ cd ~
$ git clone https://github.com/GNOME/glib.git
$ cd glib
$ git checkout 2.76.6
$ meson build --prefix=/usr
$ ninja -C build
$ sudo ninja -C build install
$ pkg-config --modversion glib-2.0
#2.76.6

# librdkafka のインストール
$ cd ~
$ git clone https://github.com/confluentinc/librdkafka.git
$ cd librdkafka
$ git checkout tags/v2.2.0
$ ./configure --enable-ssl
$ make
$ sudo make install

# DeepStream 7.1 のダウンロードとインストール
$ cd ~
$ wget --content-disposition https://api.ngc.nvidia.com/v2/resources/nvidia/deepstream/versions/7.1/files/deepstream-7.1_7.1.0-1_amd64.deb --output-document deepstream-7.1_7.1.0-1_amd64.deb
$ sudo apt-get install ./deepstream-7.1_7.1.0-1_amd64.deb

# GStreamer レジストリのクリア
$ rm -rf ~/.cache/gstreamer-1.0/

# CUDA シンボリックリンク
### sudo ln -snf /usr/local/cuda-12.6 /usr/local/cuda

# 環境変数（~/.bashrc に追記推奨）
$ vim ~/.bashrc
export PATH="/usr/local/cuda-12.6/bin${PATH:+:${PATH}}"
export LD_LIBRARY_PATH="/usr/local/cuda-12.6/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}"

$ source ~/.bashrc

# 動作確認
$ deepstream-app --version-all
#deepstream-app version 7.1.0
#DeepStreamSDK 7.1.0
#CUDA Driver Version: 13.2 # -->> NVIDIAドライバが対応している最大のCUDAバージョン
#CUDA Runtime Version: 12.6 # -->> 実際にアプリケーションがリンクして使っているCUDA Runtime
#TensorRT Version: 10.6
#cuDNN Version: 9.2
#libNVWarp360 Version: 2.0.1d3

$ cd /opt/nvidia/deepstream/deepstream-7.1/samples/configs/deepstream-app
$ deepstream-app -c source30_1080p_dec_infer-resnet_tiled_display_int8.txt

以上です！
お疲れさまでした！

0
0
0

Register as a new user and use Qiita more conveniently

  1. You get articles that match your needs
  2. You can efficiently read back useful information
  3. You can use dark theme
What you can do with signing up
Sign upLogin
0
0

Delete article

Deleted articles cannot be recovered.

Draft of this article would be also deleted.

Are you sure you want to delete this article?