NVIDIA / retinanet-examples

Fast and accurate object detection with end-to-end GPU optimization

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

[Question] about Build in ODTK

vijay180396 opened this issue · comments

os: ubuntu 18.04
arch: x86_64
using version: odtk release v0.2.0

I successfully build odtk and it also automatically added in conda environment like odtk module==0.2.0 version.
and a shared library '_C.so' file created in project folder. but it not supporting in model.py which used has module like [from _C import Engine] module import error shows.

@yashnv
can you please refer how to use _C.so file and _C.py file created in conda environment.
anyone please guide.. thank you!

without using conda or docker
and build odtk problem solved.
please refer in #38 issues.

my python version: 3.7.5
gcc, g++: 7.5.0
opencv , opencv contrib: 3.4.16 for cpp build using cmake
tensorrt: 8.4.1
nvidia-dali-cuda110: 1.14.0
tensorflow-gpu: 1.15.0
pycocotools: 2.0
numpy: 1.21.0 or latest
protobuf: 4.21.1
onnx: 1.12.0
torch: 1.8.2+cu111
torch audio: 0.8.2
torchvision: 0.9.2+cu111
apex:0.1
finally
odtk==0.2.6 installed in python3.7 site package
and _c.so problem solved