neosr is a framework for training real-world single-image super-resolution networks. wiki.


Join our Discord

news

07.12.2024 - neosr now works with TOML configuration files! Please see the templates.
06.19.2024 - neosr now supports SOTA optimization through Adan Schedule-Free and F-SAM
05.09.2024 - Release Real-PLKSR network. wiki
05.08.2024 - Release Nomos-v2 dataset. For more details, see datasets

🤝 support me

Tip

Consider supporting me on KoFi ☕ or Patreon

💻 installation

Requires Python 3.12 and CUDA =>11.8. Install latest Pytorch (=>2.3) and TorchVision (required).

Clone the repository:

git clone https://github.com/muslll/neosr
cd neosr

Then install other dependencies via pip:

pip install -e .

Alternatively, use poetry (recommended on linux):

poetry install
poetry add torch@latest torchvision@latest

Note: You must use poetry shell to enter the env after installation.

(optional) If you want to convert your models (convert.py), you need the following dependencies:

pip install onnx onnxruntime-gpu onnxconverter-common onnxsim

You can also install using poetry (recommended on linux):

poetry add onnx onnxruntime-gpu onnxconverter-common onnxsim

Please read the wiki tutorial for converting your models.

quick start

Start training by running:

python train.py -opt options.toml

Where options.toml is a configuration file. Templates can be found in options.

Tip

Please read the wiki Configuration Walkthrough for an explanation of each option.

features

Supported Archs:

arch	option
Real-ESRGAN	esrgan
SRVGGNetCompact	compact
SwinIR	swinir_small, swinir_medium
HAT	hat_s, hat_m, hat_l
OmniSR	omnisr
SRFormer	srformer_light, srformer_medium
DAT	dat_small, dat_medium, dat_2
DITN	ditn
DCTLSA	dctlsa
SPAN	span
Real-CUGAN	cugan
CRAFT	craft
SAFMN	safmn, safmn_l
RGT	rgt, rgt_s
ATD	atd, atd_light
PLKSR	plksr, plksr_tiny
RealPLKSR	realplksr, realplksr_s
DRCT	drct, drct_l, drct_s
MSDAN	msdan
SPANPlus	spanplus, spanplus_sts, spanplus_s, spanplus_st

Note

For all arch-specific parameters, read the wiki.

Under Testing

arch	option
HiT-SRF	hit_srf, hit_srf_medium, hit_srf_large
HMA	hma, hma_medium, hma_large
MAN	man, man_tiny, man_light
light-SAFMN++	light_safmnpp

Supported Discriminators:

net	option
U-Net w/ SN	unet
PatchGAN w/ SN	patchgan
EA2FPN (bespoke, based on A2-FPN	ea2fpn

Supported Optimizers:

optimizer	option
Adam	Adam or adam
AdamW	AdamW or adamw
NAdam	NAdam or nadam
Adan	Adan or adan
AdamW Win2	AdamW_Win or adamw_win
ECO strategy	eco, eco_iters
AdamW Schedule-Free	adamw_sf
Adan Schedule-Free	adan_sf
F-SAM	fsam, FSAM

Supported Losses:

loss	option
L1 Loss	L1Loss, l1_loss
L2 Loss	MSELoss, mse_loss
Huber Loss	HuberLoss, huber_loss
CHC (Clipped Huber with Cosine Similarity Loss)	chc_loss
Perceptual Loss	perceptual_opt, vgg_perceptual_loss
GAN	gan_opt, gan_loss
YCbCr Color Loss (bt601)	color_opt, color_loss
Luma Loss (CIE L*)	luma_opt luma_loss
MS-SSIM	mssim_opt mssim_loss
LDL Loss	ldl_opt, ldl_loss
Focal Frequency	ff_opt, ff_loss
DISTS	dists_opt, dists_loss
Wavelet Guided	wavelet_guided
Gradient-Weighted	gw_opt, gw_loss
Perceptual Patch Loss	perceptual_opt, patchloss, ipk

Supported Augmentations:

augmentation	option
Rotation	use_rot
Flip	use_hflip
MixUp	mixup
CutMix	cutmix
ResizeMix	resizemix
CutBlur	cutblur

Supported Models:

model	description	option
Default	Base model for SISR, supports both Generator and Discriminator	image
OTF	Builds on top of default, adding Real-ESRGAN on-the-fly degradations	otf

Supported dataset loaders:

loader	option
Paired datasets	paired
Single datasets (for inference, no GT required)	single
Real-ESRGAN on-the-fly degradation	otf

📸 datasets

As part of neosr, I have released a dataset series called Nomos. The purpose of these datasets is to distill only the best images from the academic and community datasets. A total of 14 datasets were manually reviewed and processed, including: Adobe-MIT-5k, RAISE, LSDIR, LIU4k-v2, KONIQ-10k, Nikon LL RAW, DIV8k, FFHQ, Flickr2k, ModernAnimation1080_v2, Rawsamples, SignatureEdits, Hasselblad raw samples and Unsplash.

• Nomos-v2 (recommended): contains 6000 images, multipurpose. Data distribution:

pie
  title Nomos-v2 distribution
  "Animal / fur" : 439
  "Interiors" : 280
  "Exteriors / misc" : 696
  "Architecture / geometric" : 1470
  "Drawing / painting / anime" : 1076
  "Humans" : 598
  "Mountain / Rocks" : 317
  "Text" : 102
  "Textures" : 439
  "Vegetation" : 574

Loading

• nomos_uni (recommended for lightweight networks): contains 2989 images, multipurpose. Meant to be used on lightweight networks (<800k parameters).
• hfa2k: contains 2568 anime images.

dataset download	sha256
nomosv2 (3GB)	sha256
nomosv2.lmdb (3GB)	sha256
nomosv2_lq_4x (187MB)	sha256
nomosv2_lq_4x.lmdb (187MB)	sha256
nomos_uni (1.3GB)	sha256
nomos_uni.lmdb (1.3GB)	sha256
nomos_uni_lq_4x	sha256
nomos_uni_lq_4x.lmdb	sha256
hfa2k	sha256

community datasets

Datasets made by the upscaling community. More info can be found in author's repository.

• FaceUp: Curated version of FFHQ
• SSDIR: Curated version of LSDIR.

dataset	download
@Phhofm FaceUp	GDrive (4GB)
@Phhofm SSDIR	Gdrive (4.5GB)

resources

• Training Guide from @Sirosky
• Philip's youtube channel
• OpenModelDB
• chaiNNer

📄 license and acknowledgements

Released under the Apache license. All licenses listed on license/readme. This code was originally based on BasicSR.

Thanks to victorca25/traiNNer, styler00dollar/Colab-traiNNer and timm for providing helpful insights into some problems.

Thanks to contributors @Phhofm, @Sirosky, @terrainer and @umzi2 for helping with tests and bug reporting.