Giters

Maclory / Deep-Iterative-Collaboration

Pytorch implementation of Deep Face Super-Resolution with Iterative Collaboration between Attentive Recovery and Landmark Estimation (CVPR 2020)

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

questions about testing

Darrenxc opened this issue · comments

commented

During testing, in my opinion, it will produce an 8x scale SR image w/o annotations (pkl files)?.
Can this model produce a face deblurred image without resolution changing?
Due to this model based on 8x scale, how much memory is needed to upscale a 128x128 image?

commented
  1. Yes, that's right.
  2. Since our model is trained for SR of 8x, it cannot deblur a face image without resolution changing. However, you can modify the network architecture by removing upscaling layers and train a new model for face deblurring.
  3. If the input is of 128x128, then the output will be of 1024x1024. It will cost a large amount of memory and the model of current version cannot handle such images.