This is an experimental program for real-time input video conversion with deepdream. This project is referred to this github repository.
- Windows10
- Visual Studio 2019 or 2022
- openFramewrks v0.11.2
- CUDA v11.3
- LibTorch v1.11.0
- OpenCV v4.5.5 with CUDA
LibTorch
- Download LibTorch from https://pytorch.org/get-started/locally/
- Unzip the downloaded file and copy the contents of the
include
andlib
directories to the following location.
- libtorch-win-shared-with-deps-1.11.0+cu113/include >> /libs/win/include/vs/x64/Release
- libtorch-win-shared-with-deps-1.11.0+cu113/lib >> /libs/win/lib/vs/x64/Release
- Download CUDA from https://developer.nvidia.com/cuda-toolkit-archive/
OpenCV
- Build openCV with extra CUDA module https://github.com/opencv/opencv_contrib
- Edit
addon_config.mk
to change OpenCV paths to your environment.
You don't have to follow the above method, just set up your project to use OpenCV and Libtorch.
Models
- Download models from this drive or run
/python/model_export.py
in Pytorch. - Copy
Inception_v3.pt
andspynet.pt
to themodel
directory.