[toc]

basicVSR_mge

A megengine implemention of basicVSR (CVPR 2021)

megengine	Deep learning framework created by megvii, like pytorch, tensorflow etc...	https://megengine.org.cn/
basicVSR	BasicVSR: The Search for Essential Components in Video Super-Resolution and Beyond	https://arxiv.org/abs/2012.02181

As of 2021.03.21, the official implementation（https://github.com/ckkelvinchan/BasicVSR-IconVSR has not been released. If the implementation of this repo is helpful to you, just star it, thanks!

usage

install

• Linux machine (you do not need to care about cuda version, only need NVIDIA graphics driver version greater than 418)
• python3.7
• pip3 install megengine -f https://megengine.org.cn/whl/mge.html
• pip install -r requirements.txt

dataset preparation (REDS)

• link: https://seungjunnah.github.io/Datasets/reds.html
• after unzip it ,you need to merge the training and validation dataset(like mmediting), thus total 270(240+30) clip, and remaining 30 clip for test.
• after merging, your dir should like this:
  • train
    • train_sharp
      • 000
      • ...
      • 240 (the first validation clip, thus clip 000 of validation)
      • ...
      • 269
    • train_sharp_bicubic
      • X4
        • 000
        • ...
        • 269
  • test
    • test_sharp_bicubic
      • X4
        • 000
        • ...
        • 269

Training

tutorial coming soon...

Testing (now only support REDS dataset)

get checkpoints

use our trained model(generator_module.mge), link: https://drive.google.com/drive/folders/1MXwysTaBN-3qg-iHfxl0nkv6il0EcO2K?usp=sharing

it has been trained 50 epochs on 240 clips, it's PSNR on validation dataset is 31.20(with 8 ensemble)*

test on valid dataset

find the config file: configs/restorers/BasicVSR/basicVSR_test_valid.py

change the first three line for your situation

load_path = './workdirs/epoch_xxx'                               # must have generator_module.mge file in folder epoch_xxx, xxx is digital
dataroot = "pathtoyourdataset/train/train_sharp_bicubic"
exp_name = 'basicVSR_track1_test_for_validation'                # any name you like

and then , run it:

cd xxx/basicVSR_mge
python  tools/test.py  configs/restorers/BasicVSR/basicVSR_test_valid.py --gpuids 0 -d

you can find the results in workdir

test on test dataset

same to valid, to fix the config file configs/restorers/BasicVSR/basicVSR_test_test.py first

cd xxx/basicVSR_mge
python  tools/test.py  configs/restorers/BasicVSR/basicVSR_test_test.py --gpuids 0 -d

notice: only support one gpu config for gpuids now

Results

• all output frames of test dataset produced by our model can be found here: (3000 frames, trained only on 240 training clips)

https://drive.google.com/file/d/1r1TaTAltEocXNdHY8sOgTe5b9uRB2GqA/view?usp=sharing