jiaxi-jiang / FBCNN

Official Code for ICCV 2021 paper "Towards Flexible Blind JPEG Artifacts Removal (FBCNN)"

Home Page:https://arxiv.org/abs/2109.14573

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Inference is causing Out of Memory error (even for V100)

wind-surfer opened this issue · comments

Hello,
I tried running 'main_test_fbcnn_color.py' on a real JPEG image using one 16 GB V100 but the code threw ´Out of Memory´ error. Any idea on how to use this code with large images, say 12 MPix or more?

Hi, thanks for your interest!

You can try to either run on the CPU or split the large image into smaller ones, e.g. based on the code here:https://github.com/cszn/KAIR/blob/master/utils/utils_model.py#L172

Hello,
Thank you for your prompt reply. Do you think splitting->inference->merging will produce any artifacts at the edges of the image tile?

Hi, I don't think so. I once tested large images using this method and did not see produced artifacts at the edges.

Sure. Thank you for your prompt response.