Inital Commit

.gitignore

@@ -0,0 +1,6 @@
+.idea
+*.pyc
+#fpn_inception.h5
+#fpn_mobilenet.h5
+*.h5
+*__pyc*

LICENSE

@@ -0,0 +1,58 @@
+Copyright (c) 2019, Orest Kupyn, Tetiana Martyniuk, Junru Wu and Zhangyang Wang
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+* Redistributions of source code must retain the above copyright notice, this
+  list of conditions and the following disclaimer.
+
+* Redistributions in binary form must reproduce the above copyright notice,
+  this list of conditions and the following disclaimer in the documentation
+  and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
+FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+
+
+--------------------------- LICENSE FOR DeblurGANv2 --------------------------------
+BSD License
+
+For DeblurGANv2 software
+Copyright (c) 2019, Orest Kupyn, Tetiana Martyniuk, Junru Wu and Zhangyang Wang
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+* Redistributions of source code must retain the above copyright notice, this
+  list of conditions and the following disclaimer.
+
+* Redistributions in binary form must reproduce the above copyright notice,
+  this list of conditions and the following disclaimer in the documentation
+  and/or other materials provided with the distribution.
+
+----------------------------- LICENSE FOR DCGAN --------------------------------
+BSD License
+
+For dcgan.torch software
+
+Copyright (c) 2015, Facebook, Inc. All rights reserved.
+
+Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
+
+Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
+
+Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
+
+Neither the name Facebook nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

README.md

@@ -1,2 +1,7 @@
-# DeblurGANv2_Customdataset
-Using deblur GAN on custom dataset
+# Using deblurv2 GAN on custom dataset
+
+
+
+**Credits**
+
+- [DeblurGANv2](https://github.com/VITA-Group/DeblurGANv2)

adversarial_trainer.py

@@ -0,0 +1,99 @@
+import torch
+import copy
+
+
+class GANFactory:
+    factories = {}
+
+    def __init__(self):
+        pass
+
+    def add_factory(gan_id, model_factory):
+        GANFactory.factories.put[gan_id] = model_factory
+
+    add_factory = staticmethod(add_factory)
+
+    # A Template Method:
+
+    def create_model(gan_id, net_d=None, criterion=None):
+        if gan_id not in GANFactory.factories:
+            GANFactory.factories[gan_id] = \
+                eval(gan_id + '.Factory()')
+        return GANFactory.factories[gan_id].create(net_d, criterion)
+
+    create_model = staticmethod(create_model)
+
+
+class GANTrainer(object):
+    def __init__(self, net_d, criterion):
+        self.net_d = net_d
+        self.criterion = criterion
+
+    def loss_d(self, pred, gt):
+        pass
+
+    def loss_g(self, pred, gt):
+        pass
+
+    def get_params(self):
+        pass
+
+
+class NoGAN(GANTrainer):
+    def __init__(self, net_d, criterion):
+        GANTrainer.__init__(self, net_d, criterion)
+
+    def loss_d(self, pred, gt):
+        return [0]
+
+    def loss_g(self, pred, gt):
+        return 0
+
+    def get_params(self):
+        return [torch.nn.Parameter(torch.Tensor(1))]
+
+    class Factory:
+        @staticmethod
+        def create(net_d, criterion): return NoGAN(net_d, criterion)
+
+
+class SingleGAN(GANTrainer):
+    def __init__(self, net_d, criterion):
+        GANTrainer.__init__(self, net_d, criterion)
+        self.net_d = self.net_d.cuda()
+
+    def loss_d(self, pred, gt):
+        return self.criterion(self.net_d, pred, gt)
+
+    def loss_g(self, pred, gt):
+        return self.criterion.get_g_loss(self.net_d, pred, gt)
+
+    def get_params(self):
+        return self.net_d.parameters()
+
+    class Factory:
+        @staticmethod
+        def create(net_d, criterion): return SingleGAN(net_d, criterion)
+
+
+class DoubleGAN(GANTrainer):
+    def __init__(self, net_d, criterion):
+        GANTrainer.__init__(self, net_d, criterion)
+        self.patch_d = net_d['patch'].cuda()
+        self.full_d = net_d['full'].cuda()
+        self.full_criterion = copy.deepcopy(criterion)
+
+    def loss_d(self, pred, gt):
+        return (self.criterion(self.patch_d, pred, gt) + self.full_criterion(self.full_d, pred, gt)) / 2
+
+    def loss_g(self, pred, gt):
+        return (self.criterion.get_g_loss(self.patch_d, pred, gt) + self.full_criterion.get_g_loss(self.full_d, pred,
+                                                                                                  gt)) / 2
+
+    def get_params(self):
+        return list(self.patch_d.parameters()) + list(self.full_d.parameters())
+
+    class Factory:
+        @staticmethod
+        def create(net_d, criterion): return DoubleGAN(net_d, criterion)
+

aug.py

@@ -0,0 +1,77 @@
+from typing import List
+
+import albumentations as albu
+
+
+def get_transforms(size: int, scope: str = 'geometric', crop='random'):
+    augs = {'weak': albu.Compose([albu.HorizontalFlip(),
+                                  ]),
+            'geometric': albu.OneOf([albu.HorizontalFlip(always_apply=True),
+                                     albu.ShiftScaleRotate(always_apply=True),
+                                     albu.Transpose(always_apply=True),
+                                     albu.OpticalDistortion(always_apply=True),
+                                     albu.ElasticTransform(always_apply=True),
+                                     ])
+            }
+
+    aug_fn = augs[scope]
+    crop_fn = {'random': albu.RandomCrop(size, size, always_apply=True),
+               'center': albu.CenterCrop(size, size, always_apply=True)}[crop]
+    pad = albu.PadIfNeeded(size, size)
+
+    pipeline = albu.Compose([aug_fn, pad, crop_fn], additional_targets={'target': 'image'})
+
+    def process(a, b):
+        r = pipeline(image=a, target=b)
+        return r['image'], r['target']
+
+    return process
+
+
+def get_normalize():
+    normalize = albu.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])
+    normalize = albu.Compose([normalize], additional_targets={'target': 'image'})
+
+    def process(a, b):
+        r = normalize(image=a, target=b)
+        return r['image'], r['target']
+
+    return process
+
+
+def _resolve_aug_fn(name):
+    d = {
+        'cutout': albu.Cutout,
+        'rgb_shift': albu.RGBShift,
+        'hsv_shift': albu.HueSaturationValue,
+        'motion_blur': albu.MotionBlur,
+        'median_blur': albu.MedianBlur,
+        'snow': albu.RandomSnow,
+        'shadow': albu.RandomShadow,
+        'fog': albu.RandomFog,
+        'brightness_contrast': albu.RandomBrightnessContrast,
+        'gamma': albu.RandomGamma,
+        'sun_flare': albu.RandomSunFlare,
+        'sharpen': albu.Sharpen,
+        'jpeg': albu.ImageCompression,
+        'gray': albu.ToGray,
+        'pixelize': albu.Downscale,
+        # ToDo: partial gray
+    }
+    return d[name]
+
+
+def get_corrupt_function(config: List[dict]):
+    augs = []
+    for aug_params in config:
+        name = aug_params.pop('name')
+        cls = _resolve_aug_fn(name)
+        prob = aug_params.pop('prob') if 'prob' in aug_params else .5
+        augs.append(cls(p=prob, **aug_params))
+
+    augs = albu.OneOf(augs)
+
+    def process(x):
+        return augs(image=x)['image']
+
+    return process

config/config.yaml

@@ -0,0 +1,74 @@
+%%writefile config/config.yaml
+---
+project: deblur_gan
+experiment_desc: fpn
+
+train:
+#  files_a: &FILES_A /datasets/my_dataset/**/*.jpg
+  #Absolutepath can be used
+  files_a: &FILES_A /content/Licenseplate_blur_clear_dataset/train/blur/*.jpg
+  files_b: &FILES_B /content/Licenseplate_blur_clear_dataset/train/sharp/*.jpg
+#  files_a: &FILES_A ./dataset1/blur/*.png
+#  files_b: &FILES_B ./dataset1/sharp/*.png
+  size: &SIZE 256
+  crop: random
+  preload: &PRELOAD false
+  preload_size: &PRELOAD_SIZE 0
+  bounds: [0, .9]
+  scope: geometric
+  corrupt: &CORRUPT
+    - name: cutout
+      prob: 0.5
+      num_holes: 3
+      max_h_size: 25
+      max_w_size: 25
+    - name: jpeg
+      quality_lower: 70
+      quality_upper: 90
+    - name: motion_blur
+    - name: median_blur
+    - name: gamma
+    - name: rgb_shift
+    - name: hsv_shift
+    - name: sharpen
+
+val:
+  files_a: &FILES_C /content/Licenseplate_blur_clear_dataset/valid/blur/*.jpg
+  files_b: &FILES_D /content/Licenseplate_blur_clear_dataset/valid/sharp/*.jpg
+#  files_a: &FILES_A
+#  files_b: &FILES_B
+  size: *SIZE
+  scope: geometric
+  crop: center
+  preload: *PRELOAD
+  preload_size: *PRELOAD_SIZE
+  bounds: [.9, 1]
+  corrupt: *CORRUPT
+
+phase: train
+warmup_num: 3
+model:
+  g_name: fpn_inception
+  blocks: 9
+  d_name: double_gan # may be no_gan, patch_gan, double_gan, multi_scale
+  d_layers: 3
+  content_loss: perceptual
+  adv_lambda: 0.001
+  disc_loss: wgan-gp
+  learn_residual: True
+  norm_layer: instance
+  dropout: True
+
+num_epochs: 20
+train_batches_per_epoch: 102
+val_batches_per_epoch: 100
+batch_size: 16
+image_size: [256, 256]
+
+optimizer:
+  name: adam
+  lr: 0.01
+scheduler:
+  name: linear
+  start_epoch: 20
+  min_lr: 0.0000001