Skip to content

Team <skyb> solution for the AIM2020 mobile image signal processing challenge

License

Notifications You must be signed in to change notification settings

egyptdj/skyb-aim2020-public

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Team skyb solution for the AIM2020 mobile image signal processing challenge

Publication

PyNET-CA: Enhanced PyNET with Channel Attention for End-to-End Mobile Image Signal Processing
Byung-Hoon Kim, Joonyoung Song, Jong Chul Ye, JaeHyun Baek
ECCV 2020 Workshops - Advances in Image Manipulation (AIM)

How-to

Reproduce the final results:

  1. Download the pre-trained model and extract it in the git path
  2. Run the following code with path_to_images indicating RAW images to process (add --perceptual flag for perceptual track results)
python main.py --skip_train --test_dir path_to_images
python main.py --skip_train --test_dir path_to_images --perceptual
  1. Resolved images can be found at path_to_images + '_enhanced'

Train from scratch:

  1. Download the ZRR training dataset and extract it in the data/ folder within the git path
  2. Run the following code.
python main.py

Command-line options can be listed by running the main script with -h flag.

python main.py -h

Requirements

Note: inferring with the pretrained model may not reproduce sufficient results with pytorch version over 1.4.0

  • python 3.6
  • pytorch >= 1.4.0
  • tensorboard
  • pytorch-msssim
  • IQA-pytorch
  • tqdm

Concept

PyNet-CA: Enhanced PyNet with Channel Attention for Mobile ISP

concept

Cite

@inproceedings
{
title={PyNET-CA: enhanced PyNET with channel attention for end-to-end mobile image signal processing},
author={Kim, Byung-Hoon and Song, Joonyoung and Ye, Jong Chul and Baek, JaeHyun},
booktitle={European Conference on Computer Vision},
pages={202--212},
year={2020},
organization={Springer}
}

Contact

egyptdj@yonsei.ac.kr

About

Team <skyb> solution for the AIM2020 mobile image signal processing challenge

Topics

Resources

License

Stars

Watchers

Forks

Languages