Welcome to Jetson Orin NANO Developer Kit repository
03.Torchvision
Torchvision Environment Setup
Torchvision Environment Setup1. System Information2. Installing Dependencies3. Installing Torchvision3.1. Offline Installation3.2. Online Installation3.3. Building from Source4. Verifying the InstallationReferences
Note on pip installation: Ubuntu 24.04 marks the system Python as "externally managed" according to PEP 668, so installing packages directly with pip may report
externally-managed-environment. The--break-system-packagesoption allows pip to install packages in that environment; virtual environments usually do not need this option, but it is kept here to keep the commands in this tutorial consistent.
The factory system already has the relevant environment set up, so the installation steps can be skipped.
Torchvision provides computer vision datasets, models, and image transforms for PyTorch.
1. System Information

The current environment uses PyTorch 2.8.0, Torchvision 0.23.0, and Python 3.12. Torchvision 0.23.0 corresponds to PyTorch 2.8.0.
2. Installing Dependencies
sudo apt updatesudo apt install ninja-build libwebp-dev libjpeg-dev -y3. Installing Torchvision
According to the version correspondence between PyTorch and Torchvision, PyTorch 2.8 uses Torchvision 0.23, which supports Python 3.9 through 3.13. When installing, use the version that fits Jetson ARM64, Python 3.12, and the current PyTorch.
3.1. Offline Installation
Prepare in advance the Torchvision 0.23.0 wheel file that matches the current environment, and copy it to the Jetson. Do not use the 0.20.0a0 / cp310 wheels from the old tutorials.
xxxxxxxxxxcd Downloads/python3 -m pip install --user --break-system-packages ./<Torchvision-0.23.0 ARM64 wheel file name>3.2. Online Installation
On a device with internet access, install from the PyTorch Vision v0.23.0 source. This method uses the currently installed PyTorch and does not automatically replace its dependencies:
xxxxxxxxxxpython3 -m pip install --user --break-system-packages --no-deps --no-build-isolation "git+https://github.com/pytorch/vision.git@v0.23.0"3.3. Building from Source
Download the v0.23.0 source code:
https://github.com/pytorch/vision/tree/v0.23.0
Enter the source code directory and install:
xxxxxxxxxxcd vision-0.23.0python3 -m pip install --user --break-system-packages --no-deps --no-build-isolation .4. Verifying the Installation
xxxxxxxxxxpython3 -c "import torch, torchvision; print(f'Torch: {torch.__version__}'); print(f'Torchvision: {torchvision.__version__}'); print('CUDA:', torch.cuda.is_available())"