search

KugaMaxx / cuke-emlb

A benchmark for event-based denoising.
MIT License
30 stars 4 forks source link

cmake error #4

Closed CheungBH closed 4 months ago

CheungBH commented 7 months ago

Hello. Thanks for your great work.

I wonder whether you have the plan to release the training code?

KugaMaxx commented 7 months ago

Hi, @CheungBH ! Thank you for your attention to my work. As the code is based on the implementation of Libtorch, it is difficult to replicate the training process in C++, so there is no relevant plan in the short term. :disappointed:

CheungBH commented 5 months ago

Thank you for your reply. I want to evaluate your edncnn & mplf model, but a problem occurs that: ImportError: cannot import name 'event_denoise_convolution_network' from 'modules.python' (unknown location) Are there any suggestions for solving it? I noticed that the hpp files in include/denoisors are complete, but there are only 6 .so files in module/python folder.

Screenshot 2024-01-26 at 4 55 07 PM

KugaMaxx commented 5 months ago

I think you need try:

# create folder
mkdir build && cd build

# compile with samples
CC=gcc-10 CXX=g++-10 cmake .. -DEMLB_ENABLE_PYTHON=ON -DTORCH_DIR=<path/to/libtorch>/share/cmake/Torch/

# generate library
cmake --build . --config Release

And you will see .cpython-38-x86_64-linux-gnu.so file in /modules/python. screenshot

Check if event_denoise_convolution_network.cpython-38-x86_64-linux-gnu.so and multi_layer_perceptron_filter.cpython-38-x86_64-linux-gnu.so exist, then run the code.