Problem using jetson tx2 with jetpack 4.6.3 with ultralytics

My task is to run a docker image where I will work with YOLOv8, I also need to have access to the GPU, which I organized with pytorch. I am running on a jetson tx2 with jetpack version 4.6.3 with python 3.6. To work with YOLO, I need ultralytics, and it requires python 3.7 and higher. Please help me to solve this problem. I took the base image from Nvidia l4t-ml:r32.7.1-py3 with the following packages installed in it: TensorFlow 1.15.5
PyTorch v1.10.0
torchvision v0.11.0
torchaudio v0.10.0
onnx 1.11.0
CuPy 9.2.0
numpy 1.19.5
numba 0.53.1
OpenCV 4.5.0 (with CUDA)
pandas 1.1.5
scipy 1.5.4
scikit-learn 0.24.2
JupyterLab 2.2.9
And I add CUDA settings to it and clone the YOLO repository.


We only provide the package for default python version.
So you will need to build PyTorch from the source.