Python DnCnn Exemples

Exemple #1

Fichier : outside_dncnn_ntire_config_41_4.py Projet : zyh911/image_dehazing

plot_loss_start_epoch = 1
only_validate = False  #

from visdom import Visdom
vis = Visdom(server='http://127.0.0.1', port=8097)

# =================== config for model and dataset =====================================================================
from squid.data import Photo2PhotoData
from squid.data import RandomCropPhoto2PhotoData
from squid.model import SuperviseModel
import torch
import torch.nn as nn
from squid.loss import VGGLoss
from squid.net import DnCnn

target_net = DnCnn(layer_num=20)
target_net = nn.DataParallel(target_net).cuda()

model = SuperviseModel({
    'net':
    target_net,
    'optimizer':
    torch.optim.Adam([{
        'name': 'net_params',
        'params': target_net.parameters(),
        'base_lr': 1e-4
    }],
                     betas=(0.9, 0.999),
                     weight_decay=0.0005),
    'lr_step_ratio':
    0.5,

Exemple #2

Fichier : dncnn_configv5.py Projet : zyh911/image_dehazing

only_validate = False  #

from visdom import Visdom
vis = Visdom(server='http://127.0.0.1', port=8097)


# =================== config for model and dataset =====================================================================
from squid.data import Photo2PhotoData
from squid.data import RandomCropPhoto2PhotoData
from squid.model import SuperviseModel
import torch.nn as nn
from squid.loss import VGGLoss
from squid.net import DnCnn

hr_size = 256
target_net = DnCnn(layer_num=20)
target_net.load_state_dict(checkpoint)

model = SuperviseModel({
    'net': target_net, 
    'optimizer': torch.optim.Adam([{'name':'net_params', 'params':target_net.parameters(), 'base_lr':0.00005}], betas=(0.9, 0.999), weight_decay=0.0005),
    'lr_step_ratio': 0.5,
    'lr_step_size': 5,

    'supervise':{
        'out':  {'MSE_loss': {'obj': nn.MSELoss(size_average=True),  'factor':0.1, 'weight': 1.0}}, 
    },
    'metrics': {}
     
})

Exemple #3

Fichier : outside_dncnn_ntire_config_41_3.py Projet : zyh911/image_dehazing

only_validate = False  #

from visdom import Visdom
vis = Visdom(server='http://127.0.0.1', port=8097)


# =================== config for model and dataset =====================================================================
from squid.data import Photo2PhotoData
from squid.data import RandomCropPhoto2PhotoData
from squid.model import SuperviseModel
import torch
import torch.nn as nn
from squid.loss import VGGLoss
from squid.net import DnCnn

target_net = DnCnn(layer_num=20)
target_net = nn.DataParallel(target_net).cuda()

model = SuperviseModel({
    'net': target_net, 
    'optimizer': torch.optim.Adam([{'name':'net_params', 'params':target_net.parameters(), 'base_lr':1e-4}], betas=(0.9, 0.999), weight_decay=0.0005),
    'lr_step_ratio': 0.5,
    'lr_step_size': 20,

    'supervise':{
        'out':  {'MSE_loss': {'obj': nn.MSELoss(size_average=True),  'factor':1.0, 'weight':1.0}}, 
    },
    'metrics': {}
     
})

Exemple #4

Fichier : inference_config.py Projet : zyh911/image_dehazing

# inference config file
# Created by zyh in Meitu.

from squid.net import DnCnn
from squid.net import AODNet
from squid.net import DnCnn_AOD

test_snapshot_path = '/root/zyh3/train_tasks/dncnn_configv5/models/snapshot_12_G_model'

# target_net = AODNet()
target_net = DnCnn(layer_num=20)
# target_net = DnCnn_AOD()

test_input_dir = '/root/zyh3/SOTS/SOTS/indoor/nyuhaze500/hazy'
# test_input_dir = '/root/zyh3/IndoorTrain/IndoorTrainHazy'
TEST_OUT_FOLDER = '/root/zyh3/SOTS_dncnn_20layer_out'
# TEST_OUT_FOLDER = '/root/zyh3/IndoorTrain/IndoorTrainHazy_out'
GPU_ID = 0

vis = None

Exemple #5

only_validate = False  #

from visdom import Visdom
vis = Visdom(server='http://127.0.0.1', port=8097)

# =================== config for model and dataset =====================================================================
from squid.data import Photo2PhotoData
from squid.data import RandomCropPhoto2PhotoData
from squid.model import SuperviseModel
import torch
import torch.nn as nn
from squid.loss import VGGLoss
from squid.net import DnCnn

hr_size = (407, 541)
target_net = DnCnn()

model = SuperviseModel({
    'net':
    target_net,
    'optimizer':
    torch.optim.Adam([{
        'name': 'net_params',
        'params': target_net.parameters(),
        'base_lr': 2e-4
    }],
                     betas=(0.9, 0.999),
                     weight_decay=0.0005),
    'lr_step_ratio':
    0.5,
    'lr_step_size':

Exemple #6

plot_loss_start_epoch = 1
only_validate = False  #

from visdom import Visdom
vis = Visdom(server='http://127.0.0.1', port=8097)

# =================== config for model and dataset =====================================================================
from squid.data import Photo2PhotoData
from squid.data import RandomCropPhoto2PhotoData
from squid.model import SuperviseModel
import torch.nn as nn
from squid.loss import VGGLoss
from squid.net import DnCnn

hr_size = (407, 541)
target_net = DnCnn()
target_net.load_state_dict(checkpoint)

model = SuperviseModel({
    'net':
    target_net,
    'optimizer':
    torch.optim.Adam([{
        'name': 'net_params',
        'params': target_net.parameters(),
        'base_lr': 1e-4
    }],
                     betas=(0.9, 0.999),
                     weight_decay=0.0005),
    'lr_step_ratio':
    0.5,