That is just a first draft, with some dirty hacks. Among other, I was not able to get dynamically the input/output tensor shapes for I/O layers, I also removed compatibility for INT8 (just need to replace std::filesystem::directory_iterator), and not sure if the default seed for random number is always the same in std::default_random_engine.
Let me know how you want to proceed, from there I see 3 options:
Changes in another branch feat/jetson-tx2 (as I did on my fork). I recommend this and I would need you to create this branch on your repo so I can make a PR.
Changes living in main branch. I don't recommend this to avoid polluting your original code but that could be possible to have everything at the same place with a bunch of #ifdef depending on cuda version , c++ etc...
No mention to this work on your repo, even if I think it could benefit the community.
Hi, and thanks for your work.
I updated your code to make it work on a jetson-TX2 which is compatible with:
https://github.com/ltetrel/tensorrt-cpp-api/tree/feat/jetson-tx2
That is just a first draft, with some dirty hacks. Among other, I was not able to get dynamically the input/output tensor shapes for I/O layers, I also removed compatibility for INT8 (just need to replace
std::filesystem::directory_iterator
), and not sure if the default seed for random number is always the same instd::default_random_engine
.Let me know how you want to proceed, from there I see 3 options:
feat/jetson-tx2
(as I did on my fork). I recommend this and I would need you to create this branch on your repo so I can make a PR.main
branch. I don't recommend this to avoid polluting your original code but that could be possible to have everything at the same place with a bunch of#ifdef
depending on cuda version , c++ etc...See also: https://github.com/cyrusbehr/YOLOv8-TensorRT-CPP/issues/22
Let me know,