Open lb-hit opened 2 years ago
@lvhan028
You can refer to this guide: https://mmdeploy.readthedocs.io/en/latest/01-how-to-build/jetsons.html
@lvhan028 @tpoisonooo Hello, I have visited that link before. The link seems to say only how to install mmdeploy on jetson and the pre-installation preparation (JetPack SDK, Conda, PyTorch, CMake, TensorRT, MMCV, ONNX). And in another link (https://mmdeploy.readthedocs.io/zh_CN/latest/get_started.html), it seems to say only how to install MMDeploy in workstation. So after reading it, I don't really know how the pipeline works.
get_started
is limited to getting started on PC or Server.
Regarding jetson, https://mmdeploy.readthedocs.io/en/latest/01-how-to-build/jetsons.html is the actual getting started.
It presents:
@lvhan028 Hello, I would like to ask you again.
@lvhan028 Hello, I am getting an error that the version does not meet the requirements (cmake>=3.1) when installing mmcv on Jetson NX. How should I fix this? (Already installed cmake v3.23.1, follow the link https://mmdeploy.readthedocs.io/en/latest/01-how-to-build/jetsons.html )
@lvhan028 Hello, I got another error when installing onnx on Jetson NX, how can I fix it?
@lvhan028 Hello, I got another error when installing h5py and pycuda on Jetson NX, how can I fix it? Thanks!
For example, from a model trained on a workstation to a model deployed on jetson.