Qunliang Xing

Support c=1 images in `FramesToTensor`

2

Hi @ckkelvinchan When I used `FramesToTensor` for my gray-scale images, I met a bug. And I solve it in this PR. In specific, I want to train and test my...

Hi, I am wondering that is there a need to restore the dataloader condition. Every time we restore the model, we only restore the model parameters, optimizer, and scheduler. The...

Why not use tanh at the output of ESRGAN?

1

Dear author I notice that a lot of GAN works such as SRGAN are using tanh activation at the output side. Tanh can ensure the output to range from -1...

Code: https://github.com/open-mmlab/mmediting/blob/e1d488323096c0850a08919ffdb03e0139ca66d7/mmedit/models/backbones/sr_backbones/rdn.py#L54 The output channel number of `self.lff` is `channel_growth`. However, the input channel number of `x` is `in_channels`. In the `forward` function, the residual learning may lead to error....

Hi, thanks for your great work! I am wondering if this plugin supports listing and deleting empty folders. It would be helpful if it can handle both files and folders.