Is there any Spatial Transformer Layer kind of a thing in pytorch? I could find
TransformerLayer in Lasagne which is the STN layer implementation.
If there is any example of STN with
grid_sample as mentioned below, it would be of great help.
Thanks for the links.
I would like to ask if there any example to depict the combination of the the
grid_sample along with the localization network (which is essentially a set of conv layers ending up to regress the 6 parameters of the transformation)?
I just try an example without adding localization network but a pre-defined parameters for transformation. I think it’s straight forward to be modified.
This file has been truncated.
"## Pytorch spatial tranformer Example\n",
"See also https://github.com/fxia22/stn.pytorch\n",
"Note that torch.nn.functional.affine_grid and torch.nn.functional.grid_sample only in Pytorch v0.2.0 or above"
@YongyiTang92 For Pytorch v0.2.0, is there a direct link that we can download the builded whl file, or we have to build from the source file?
Only source file is the only way to install v0.2.0 currently.