独自のデータセットでReal-ESRGANをファインチューニングする。

git clone https://github.com/xinntao/Real-ESRGAN.git
cd Real-ESRGAN
mkdir venv
python -m venv venv
source venv/bin/activate
pip install basicsr
pip install facexlib
pip install gfpgan
pip install -r requirements.txt
python setup.py develop
deactivate

python scripts/generate_meta_info_pairdata.py --input datasets/DF2K/DIV2K_train_HR_sub datasets/DF2K/DIV2K_train_LR_bicubic_X4_sub --meta_info datasets/DF2K/meta_info/meta_info_DIV2K_sub_pair.txt

wget https://github.com/xinntao/Real-ESRGAN/releases/download/v0.1.0/RealESRGAN_x4plus.pth -P experiments/pretrained_models

wget https://github.com/xinntao/Real-ESRGAN/releases/download/v0.2.2.3/RealESRGAN_x4plus_netD.pth -P experiments/pretrained_models

train:
    name: DIV2K
    type: RealESRGANPairedDataset
    dataroot_gt: datasets/DF2K  # modify to the root path of your folder
    dataroot_lq: datasets/DF2K  # modify to the root path of your folder
    meta_info: datasets/DF2K/meta_info/meta_info_DIV2K_sub_pair.txt  # modify to your own generate meta info txt
    io_backend:
        type: disk

pretrain_network_g: experiments/pretrained_models/RealESRNet_x4plus.pth

pretrain_network_g: experiments/pretrained_models/RealESRGAN_x4plus.pth

ModuleNotFoundError: No module named 'torchvision.transforms.functional_tensor'

from torchvision.transforms.functional_tensor import rgb_to_grayscale

from torchvision.transforms.functional import rgb_to_grayscale

python realesrgan/train.py -opt options/finetune_realesrgan_x4plus_pairdata.yml --auto_resume

!pip install basicsr
!pip install facexlib
!pip install gfpgan
!pip install torch==2.0.1 torchvision==0.15.2 --extra-index-url https://download.pytorch.org/whl/cu118
!pip install numpy==1.26.4
!pip install -r requirements.txt
!python setup.py develop

basicsr>=1.4.2
facexlib>=0.2.5
gfpgan>=1.3.5
#numpy
opencv-python
Pillow
#torch>=1.7
#torchvision
tqdm

import cv2
import numpy as np
import os
from pathlib import Path

# Real-ESRGANのファインチューニング用学習データセット作成


gray_indir = 'X:/dataset/gray/'
gray_outdir = 'X:/dataset/gray_tiles/'
os.makedirs(gray_outdir, exist_ok=True)

# トリミングサイズ
gray_tile_h, gray_tile_w = 256, 256
count = 1
for filename in os.listdir(gray_indir):
    img = cv2.imread(os.path.join(gray_indir, filename), 0)
    h, w = img.shape[:2]
    count = 0
    print(f"{filename} - size: {w}x{h}")
    for y in range(0, h - gray_tile_h + 1, gray_tile_h):
        for x in range(0, w - gray_tile_w + 1, gray_tile_w):
            print(f"{y}, {y+gray_tile_h}, {x}, {x+gray_tile_h}")
            tile = img[y:y+gray_tile_h, x:x+gray_tile_w]
            out_filename = f"{filename}-{count}.png"
            cv2.imwrite(os.path.join(gray_outdir, out_filename), tile)
            count += 1

tone_indir = 'X:/dataset/tone/'
tone_outdir = 'X:/dataset/tone_tiles/'
os.makedirs(tone_outdir, exist_ok=True)

# トリミングサイズ
tile_h, tile_w = 64, 64
count = 1
for filename in os.listdir(tone_indir):
    img = cv2.imread(os.path.join(tone_indir, filename), 0)
    h, w = img.shape[:2]
    count = 0
    print(f"{filename} - size: {w}x{h}")
    for y in range(0, h - tile_h + 1, tile_h):
        for x in range(0, w - tile_w + 1, tile_w):
            print(f"{y}, {y+tile_h}, {x}, {x+tile_w}")
            tile = img[y:y+tile_h, x:x+tile_w]
            out_filename = f"{filename}-{count}.png"
            cv2.imwrite(os.path.join(tone_outdir, out_filename), tile)
            count += 1

# general settings
name: finetune_RealESRGANx4plus_400k_pairdata
model_type: RealESRGANModel
scale: 4
num_gpu: auto
manual_seed: 0

# USM the ground-truth
l1_gt_usm: True
percep_gt_usm: True
gan_gt_usm: False

high_order_degradation: False # do not use the high-order degradation generation process

# dataset and data loader settings
datasets:
  train:
    name: DIV2K
    type: RealESRGANPairedDataset
    dataroot_gt: datasets/DF2K
    dataroot_lq: datasets/DF2K
    meta_info: datasets/DF2K/meta_info/meta_info_DIV2K_sub_pair.txt
    io_backend:
      type: disk

    gt_size: 256
    use_hflip: True
    use_rot: False

    # data loader
    use_shuffle: true
    num_worker_per_gpu: 5
#batch_size_per_gpu: 12
    batch_size_per_gpu: 17
    dataset_enlarge_ratio: 1
    prefetch_mode: ~

  # Uncomment these for validation
  # val:
  #   name: validation
  #   type: PairedImageDataset
  #   dataroot_gt: path_to_gt
  #   dataroot_lq: path_to_lq
  #   io_backend:
  #     type: disk

# network structures
network_g:
  type: RRDBNet
  scale: 4
  num_in_ch: 3
  num_out_ch: 3
  num_feat: 64
  num_block: 23
  num_grow_ch: 32

network_d:
  type: UNetDiscriminatorSN
  num_in_ch: 3
  num_feat: 64
  skip_connection: True

# path
path:
  # use the pre-trained Real-ESRNet model
  pretrain_network_g: experiments/pretrained_models/RealESRGAN_x4plus.pth
  param_key_g: params_ema
  strict_load_g: true
  pretrain_network_d: experiments/pretrained_models/RealESRGAN_x4plus_netD.pth
  param_key_d: params
  strict_load_d: true
  resume_state: ~

# training settings
train:
  ema_decay: 0.999
  optim_g:
    type: Adam
    lr: !!float 1e-4
    weight_decay: 0
    betas: [0.9, 0.99]
  optim_d:
    type: Adam
    lr: !!float 1e-4
    weight_decay: 0
    betas: [0.9, 0.99]

  scheduler:
    type: MultiStepLR
    milestones: [400000]
    gamma: 0.5

  total_iter: 400000
  warmup_iter: -1  # no warm up

  # losses
  pixel_opt:
    type: L1Loss
    loss_weight: 1.0
    reduction: mean
  # perceptual loss (content and style losses)
  perceptual_opt:
    type: PerceptualLoss
    layer_weights:
      # before relu
      'conv1_2': 0.1
      'conv2_2': 0.1
      'conv3_4': 1
      'conv4_4': 1
      'conv5_4': 1
    vgg_type: vgg19
    use_input_norm: true
    perceptual_weight: !!float 1.0
    style_weight: 0
    range_norm: false
    criterion: l1
  # gan loss
  gan_opt:
    type: GANLoss
    gan_type: vanilla
    real_label_val: 1.0
    fake_label_val: 0.0
    loss_weight: !!float 1e-1

  net_d_iters: 1
  net_d_init_iters: 0

# Uncomment these for validation
# validation settings
# val:
#   val_freq: !!float 5e3
#   save_img: True

#   metrics:
#     psnr: # metric name
#       type: calculate_psnr
#       crop_border: 4
#       test_y_channel: false

# logging settings
logger:
  print_freq: 100
  save_checkpoint_freq: !!float 5e3
  use_tb_logger: true
  wandb:
    project: ~
    resume_id: ~

# dist training settings
dist_params:
  backend: nccl
  port: 29500

cp /mnt/x/dataset/gray_tiles/* datasets/DF2K/gray/

cp /mnt/x/dataset/tone_tiles/* datasets/DF2K/tone/

python3 scripts/generate_meta_info_pairdata.py --input datasets/DF2K/gray datasets/DF2K/tone --meta_info datasets/DF2K/meta_info/meta_info_DIV2K_sub_pair.txt

#!/bin/bash
cd ~
venvFile=`pwd`'/git/Real-ESRGAN/venv/bin/activate'
dataRoot='/mnt/z/work'
inputDir=$dataRoot'/in'
outputDir=$dataRoot'/out'
logfile=$dataRoot'/out/Real-ESRGAN.log'

cd $outputDir
if [ `ls -U1 | wc -l` -ne 0 ]; then
    rm -rf *
fi;
cd $inputDir
if [ `ls -U1 | wc -l` -eq 0 ]; then
    echo 'Not File';
    exit 1
fi;
cd ~/git/Real-ESRGAN
source $venvFile
echo '*** Real-ESRGAN Start ***' >> $logfile
date >> $logfile
python inference_realesrgan.py -n RealESRGAN_x4plus --model_path experiments/finetune_RealESRGANx4plus_400k_pairdata/models/net_g_大きな値が直近.pth --suffix '' -i $inputDir -o $outputDir >> $logfile
date >> $logfile
echo '*** Real-ESRGAN End ***' >> $logfile
deactivate
cd $inputDir
if [ `ls -U1 | wc -l` -ne 0 ]; then
    rm *
fi;