Setting up the YOLO Environment

1. System Information

2. Downloading Files

This example uses ultralytics-v8.4.21.

cd ~/ 
git clone https://github.com/ultralytics/ultralytics.git -b v8.4.21 --depth=1

If you cannot clone the repository due to network issues, you can download the compressed file from the following website and then extract it.

xxxxxxxxxx
https://github.com/ultralytics/ultralytics/releases/tag/v8.4.21

3. Resolving Dependency Libraries

xxxxxxxxxx
sudo pip3 install -U protobuf --ignore-installed --break-system-packages

xxxxxxxxxx
sudo pip3 install -U packaging --ignore-installed --break-system-packages

xxxxxxxxxx
sudo pip3 install -U attrs --ignore-installed --break-system-packages

4. Installing Ultralytics

xxxxxxxxxx
cd ~/ultralytics
sudo pip3 install ".[export]" --break-system-packages -i https://pypi.tuna.tsinghua.edu.cn/simple

5. Configuring GPU Acceleration

Since we have already installed torch 2.10.0 and torchvision 0.25.0 in the previous tutorial, there is no need to run the torch and torchvision installation commands here; we only need to install the remaining packages.

torch

torchvision

onnxruntime-gpu

xxxxxxxxxx
sudo pip3 install https://github.com/ultralytics/assets/releases/download/v0.0.0/onnxruntime_gpu-1.24.0-cp312-cp312-linux_aarch64.whl --break-system-packages

6. Verify Installation

Verify Ultralytics

xxxxxxxxxx
python3 -c "import ultralytics; print(ultralytics.__version__)"

Verify Torch

xxxxxxxxxx
python3 -c "import torch; print(torch.__version__); print(torch.cuda.is_available())"

Verify Torchvision

xxxxxxxxxx
python3 -c "import torchvision; print(torchvision.__version__)"

Verify NumPy

xxxxxxxxxx
python3 -c "import numpy; print(numpy.__version__)"

Verify onnxruntime-gpu

xxxxxxxxxx
pip3 list | grep onnx

References

https://docs.ultralytics.com/guides/nvidia-jetson/