search

saikatdutta / Stacked_DMSHN_bokeh

Official Repository for our CVPRW (MAI'21) paper.
MIT License
22 stars 8 forks source link

Questions about runtime？ #3

Closed C-water closed 3 years ago

C-water commented 3 years ago

Hi,

I tried to test runtime, but I could not find detailed measurement details in the paper. I have some questions :

Q1. Is the runtime average for VAL294? Do I need to remove the running time of the first slide?

Q2. What piece of code does the runtime measure? Does it include image preprocessing? Or only calculate the time of model. predict? start = time.time() bok_pred = bokehnet(input_image) end = time.time() - start print(end)

Thank you for your help.

saikatdutta commented 3 years ago

Hi,

  1. Yes, it's averaged over Val294 set. Generally, removing first few slides won't matter much since we are taking average on ~300 samples.
  2. Your pseudocode is correct. We use only the GPU time and exclude any I/O and preprocessing time. Don't forget to use torch.no_grad().
C-water commented 3 years ago

Hi,

I tested the DMSHN average runtime, which was about 0.017s. Very fast !

Thank you.