wangruixing
wangruixing
I also meet this problem, I use VOC2012, and pretrained model..
The hdrnet_ops.so file should be made in your local environment, then it can run correctly, please make sure your environment is correct, cuda8.0, tensorflow1.1, python2.7. https://github.com/wangruixing/DeepUPE/issues/22 The issue may help...
> 东西不全就别开源，害人么？！ 这个是需要在本地上make出来的，每个人环境路径不同，直接给你一个编译出来的，你也run不了。这里面有makefile。需要你自己去make
You mean encoder network and feature extractor? Actually these don't affect the final result. The key idea is use this network to learn an illumination map.
Because of the copyright problem, we still need some time to communicate with collaborate company.
Is the testing data from our link? At first, you should use adobe5k default option to process the data, and try use our matlab psnr code and SSIM code from...
Have you run the "make" command?
It seems you are trying to run the project without GPU, what's your tensorflow version?
> I have installed tf with the requirement.txt in your package, and the version is 1.1.0 Can you open the python and try to print the tf version?
https://github.com/mgharbi/hdrnet_legacy/issues/5 The issue may help you.