search

swz30 / MIRNetv2

[TPAMI 2022] Learning Enriched Features for Fast Image Restoration and Enhancement. Results on Defocus Deblurring, Denoising, Super-resolution, and image enhancement
Other
416 stars 57 forks source link

question about RCBs number of each MRB #3

Closed RuiWang-coder closed 2 years ago

RuiWang-coder commented 2 years ago

hello my friends, your work was excellent, but i have a question: why you use 2 RCBs of each MRB in paper? In paper's Ablation Studies(table 11), 3 RCBs have better result than 2 RCBs.

swz30 commented 2 years ago

Hi,

Since our goal was to build a faster variant of MIRNet, we chose to have better computational efficiency (speed, params, etc.) at a loss of some accuracy.

RuiWang-coder commented 2 years ago

ok i see! Thanks for your quick reply~~~

jiujiaolulule commented 2 years ago

Hi，my dear friends! Could you tell me what the SIDD Datasets you choose? SIDD-Medium Dataset or SIDD-Full Dataset、Raw-RGB、sRGB? I'm looking forward to your reply!

RuiWang-coder commented 2 years ago

sorry dude, i didn't use the SIDD Datasets yet,i thought may be you should create a new issue to ask the owner?

swz30 commented 2 years ago

We use SIDD-Medium dataset. Here is the direct Google Drive link to download the training images https://drive.google.com/file/d/1UHjWZzLPGweA9ZczmV8lFSRcIxqiOVJw/view?usp=sharing

jiujiaolulule commented 2 years ago

oh! thank you very much! my dear friends!