ARM: Any-Time Super-Resolution Method (Paper Link) ![](https://camo.githubusercontent.com/7bce77b609203a96227b5d4345d395a285baae9063a433c4f7c63ed9b48aa264/68747470733a2f2f76697369746f722d62616467652e676c697463682e6d652f62616467653f706167655f69643d6368656e626f6e672e41524d2d4e6574)
Pytorch implementation of ARM-Net in ECCV 2022.
- Python 3.6
- Pytorch 1.7
Run the command python img/plot_psnr_flops.py
to get the above figure.
cd data_scripts
python extract_subimages_test.py
python data_augmentation.py
python generate_mod_LR_bic.py
python extract_subimages_train.py
# train_us_fsrcnn.yml
is_train: True
# train_us_carn.yml
is_train: True
# train_us_srresnet.yml
is_train: True
python main.py -opt options/train_us_fsrcnn.yml
python main.py -opt options/train_us_carn.yml
python main.py -opt options/train_us_srresnet.yml
# train_us_fsrcnn.yml
path:
pretrain_model_G: ckpt/arm-fsrcnn.pth
resume_state: ckpt/arm-fsrcnn.state
is_train: False
is_test: True
# train_us_carn.yml
path:
pretrain_model_G: ckpt/arm-carn.pth
resume_state: ckpt/arm-carn.state
is_train: False
is_test: True
# train_us_srresnet.yml
path:
pretrain_model_G: ckpt/arm-srresnet.pth
resume_state: ckpt/arm-srresnet.state
is_train: False
is_test: True
python main.py -opt options/train_us_fsrcnn.yml
python main.py -opt options/train_us_carn.yml
python main.py -opt options/train_us_srresnet.yml