Support for variable sized input images?

[shashankvasisht]

Hi Vinthony,

I saw that you are resizing the input image to 256x256. Is there any support for variable size inputs?

[vinthony]

hi, thanks for your attention.
Actually, our model is fully convolutional which can be tested in any resolution directly.
Currently, 256x256 is just a good choice for evaluation.