Rui Xu
Rui Xu
Hi @RexBarker , about the problem in mask processing, I think it is related to the saved type and how to deal with the masks. You may customize the pre-processing...
Which watermark? You mean the watermark in the demo video?
I just create the watermark mask by myself and multiply the mask with original images.
Could you provide much more information?
Many factors have influence on the speed and I may provide a general optimization on the speed by using more fask [flow warping](https://mmcv.readthedocs.io/en/latest/api.html?highlight=flow_warp#mmcv.video.flow_warp) in our MMCV toolkit. For run-time speed,...
Sorry for the late reply. We do not plan to write our method in tensorflow.
Hi @6Holmes , you can just try this command to promote the file's permission: ``` chmod +x ./make.sh ```
Have you checked the version of PyTorch? And here is another [branch](https://github.com/nbei/Deep-Flow-Guided-Video-Inpainting/pull/63) without installation of flownet2.
I suggest that you can just compile the correlation layer and then see whether any error information will be reported.