Cery D

> > It init's the driver, then spawns a child process > > According to Nvidia, you may not `fork()` after intializing CUDA. you can look for this issure. https://devtalk.nvidia.com/default/topic/1052699/tensorrt/how-to-implement-tensorrt-as-an-inference-server-/post/5343685/#reply...

> _No description provided._ no we use the originial model40 file

> But the data you provided in Google Cloud seems preprocessed. it is orignized by the pytorch-graph. you can also made by yourself

yes , currently we conduct some facial paralysis segment results. and also can use it in human segmentation

maybe the pytorch was update，this repo is my master degree paper，so there maybe some update with pytorch. you should update the code yourself

> As far as I understand, you want to create `.ply`, `.obj` or `.txt` files from `torch_geometric.Data` objects? We are in the process of creating `torch_geometric.io`, but this functionality is...

thank for your watching our research, this time the papar have not published yet. but it will coming soon i add some readme in the repo. May be you can...