Add Air Change Dataset

README.md

@@ -13,14 +13,14 @@ as the [official repo](https://github.com/rcdaudt/fully_convolutional_change_det
 ```bash
 # The network definition scripts are from the original repo
 git clone --recurse-submodules git@github.com:Bobholamovic/FCN-CD-PyTorch.git
+cd FCN-CD-PyTorch
+mkdir exp
+cd src
 ```
 
 For training, try
 
 ```bash
-# In the root directory of this repository
-mkdir exp
-cd src
 python train.py train --exp-config ../config_base.yaml
 ```
 

config_base.yaml

@@ -3,23 +3,23 @@
 
 # Data
 # Common
-dataset: OSCD
-crop_size: 256
+dataset: AC_Tiszadob
+crop_size: 112
 num_workers: 1
-repeats: 1000
+repeats: 3200
 
 
 # Optimizer
-optimizer: Adam
+optimizer: SGD
 lr: 0.001
 lr_mode: const
-weight_decay: 0.0001
+weight_decay: 0.0005
 step: 2
 
 
 # Training related
 batch_size: 32
-num_epochs: 40
+num_epochs: 10
 resume: ''
 load_optim: True
 anew: False
@@ -48,4 +48,4 @@ weights:
 
 # Model
 model: siamunet_conc
-num_feats_in: 13
+num_feats_in: 3
\ No newline at end of file

src/constants.py

@@ -3,7 +3,7 @@
 
 # Dataset directories
 IMDB_OSCD = '~/Datasets/OSCDDataset/'
-IMDB_AC = '~/Datasets/SZTAKI_AirChange_Benchmark/'
+IMDB_AirChange = '~/Datasets/SZTAKI_AirChange_Benchmark/'
 
 # Checkpoint templates
 CKP_LATEST = 'checkpoint_latest.pth'

src/core/factories.py

@@ -109,7 +109,7 @@ def single_model_factory(model_name, C):
 
 
 def single_optim_factory(optim_name, params, C):
-    name = optim_name.upper()
+    name = optim_name.strip().upper()
     if name == 'ADAM':
         return torch.optim.Adam(
             params, 
@@ -117,6 +117,13 @@ def single_optim_factory(optim_name, params, C):
             lr=C.lr,
             weight_decay=C.weight_decay
         )
+    elif name == 'SGD':
+        return torch.optim.SGD(
+            params, 
+            lr=C.lr,
+            momentum=0.9,
+            weight_decay=C.weight_decay
+        )
     else:
         raise NotImplementedError("{} is not a supported optimizer type".format(optim_name))
 
@@ -137,7 +144,8 @@ def single_critn_factory(critn_name, C):
 
 def single_train_ds_factory(ds_name, C):
     from data.augmentation import Compose, Crop, Flip
-    module = _import_module('data', ds_name.strip())
+    ds_name = ds_name.strip()
+    module = _import_module('data', ds_name)
     dataset = getattr(module, ds_name+'Dataset')
     configs = dict(
         phase='train', 
@@ -150,10 +158,10 @@ def single_train_ds_factory(ds_name, C):
                 root = constants.IMDB_OSCD
             )
         )
-    elif ds_name == 'AC':
+    elif ds_name.startswith('AC'):
         configs.update(
             dict(
-                root = constants.IMDB_AC
+                root = constants.IMDB_AirChange
             )
         )
     else:
@@ -171,11 +179,13 @@ def single_train_ds_factory(ds_name, C):
 
 
 def single_val_ds_factory(ds_name, C):
-    module = _import_module('data', ds_name.strip())
+    ds_name = ds_name.strip()
+    module = _import_module('data', ds_name)
     dataset = getattr(module, ds_name+'Dataset')
     configs = dict(
         phase='val', 
-        transforms=(None, None, None)
+        transforms=(None, None, None),
+        repeats=1
     )
     if ds_name == 'OSCD':
         configs.update(
@@ -183,7 +193,7 @@ def single_val_ds_factory(ds_name, C):
                 root = constants.IMDB_OSCD
             )
         )
-    elif ds_name == 'AC':
+    elif ds_name.startswith('AC'):
         configs.update(
             dict(
                 root = constants.IMDB_AirChange
@@ -246,4 +256,4 @@ def data_factory(dataset_names, phase, C):
 def metric_factory(metric_names, C):
     from utils import metrics
     name_list = _parse_input_names(metric_names)
-    return [getattr(metrics, name)() for name in name_list]
+    return [getattr(metrics, name.strip())() for name in name_list]

src/core/trainers.py

@@ -267,6 +267,6 @@ class CDTrainer(Trainer):
                 self.logger.dump(desc)
                     
                 if store:
-                    self.save_image(name[0], CM.squeeze(-1), epoch)
+                    self.save_image(name[0], (CM*255).squeeze(-1), epoch)
 
         return self.metrics[0].avg if len(self.metrics) > 0 else max(1.0 - losses.avg, self._init_max_acc)
\ No newline at end of file

src/data/AC_Szada.py

+from ._AirChange import _AirChangeDataset
+
+
+class AC_SzadaDataset(_AirChangeDataset):
+    def __init__(
+        self, 
+        root, phase='train', 
+        transforms=(None, None, None), 
+        repeats=1
+    ):
+        super().__init__(root, phase, transforms, repeats)
+
+    @property
+    def LOCATION(self):
+        return 'Szada'
+
+    @property
+    def TEST_SAMPLE_IDS(self):
+        return (0,)
+        
+    @property
+    def N_PAIRS(self):
+        return 7
\ No newline at end of file

src/data/AC_Tiszadob.py

+from ._AirChange import _AirChangeDataset
+
+
+class AC_TiszadobDataset(_AirChangeDataset):
+    def __init__(
+        self, 
+        root, phase='train', 
+        transforms=(None, None, None), 
+        repeats=1
+    ):
+        super().__init__(root, phase, transforms, repeats)
+
+    @property
+    def LOCATION(self):
+        return 'Tiszadob'
+
+    @property
+    def TEST_SAMPLE_IDS(self):
+        return (2,)
+
+    @property
+    def N_PAIRS(self):
+        return 5
\ No newline at end of file

src/data/OSCD.py

@@ -33,7 +33,7 @@ class OSCDDataset(CDDataset):
         with open(txt_file, 'r') as f:
             cities = [city.strip() for city in f.read().strip().split(',')]
         if self.phase == 'train':
-            # For training, use the first 10 pairs
+            # For training, use the first 11 pairs
             cities = cities[:-3]
         else:
             # For validation, use the remaining 3 pairs

src/data/_AirChange.py

+import abc
+from os.path import join, basename
+from multiprocessing import Manager
+
+import numpy as np
+
+from . import CDDataset
+from .common import default_loader
+from .augmentation import Crop
+
+
+class _AirChangeDataset(CDDataset):
+    def __init__(
+        self, 
+        root, phase='train', 
+        transforms=(None, None, None), 
+        repeats=1
+    ):
+        super().__init__(root, phase, transforms, repeats)
+        self.cropper = Crop(bounds=(0, 0, 748, 448))
+
+        self._manager = Manager()
+        sync_list = self._manager.list
+        self.images = sync_list([sync_list([None]*self.N_PAIRS), sync_list([None]*self.N_PAIRS)])
+        self.labels = sync_list([None]*self.N_PAIRS)
+
+    @property
+    @abc.abstractmethod
+    def LOCATION(self):
+        return ''
+
+    @property
+    @abc.abstractmethod
+    def TEST_SAMPLE_IDS(self):
+        return ()
+
+    @property
+    @abc.abstractmethod
+    def N_PAIRS(self):
+        return 0
+
+    def _read_file_paths(self):
+        if self.phase == 'train':
+            sample_ids = range(self.N_PAIRS)
+            t1_list = ['-'.join([self.LOCATION,str(i),'0.bmp']) for i in sample_ids if i not in self.TEST_SAMPLE_IDS]
+            t2_list = ['-'.join([self.LOCATION,str(i),'1.bmp']) for i in sample_ids if i not in self.TEST_SAMPLE_IDS]
+            label_list = ['-'.join([self.LOCATION,str(i),'cm.bmp']) for i in sample_ids if i not in self.TEST_SAMPLE_IDS]
+        else:
+            t1_list = ['-'.join([self.LOCATION,str(i),'0.bmp']) for i in self.TEST_SAMPLE_IDS]
+            t2_list = ['-'.join([self.LOCATION,str(i),'1.bmp']) for i in self.TEST_SAMPLE_IDS]
+            label_list = ['-'.join([self.LOCATION,str(i),'cm.bmp']) for i in self.TEST_SAMPLE_IDS]
+
+        return t1_list, t2_list, label_list
+
+    def fetch_image(self, image_name):
+        _, i, t = image_name.split('-')
+        i, t = int(i), int(t[:-4])
+        if self.images[t][i] is None:
+            image = self._bmp_loader(join(self.root, self.LOCATION, str(i+1), 'im'+str(t+1)))
+            self.images[t][i] = image if self.phase == 'train' else self.cropper(image)
+        return self.images[t][i]
+
+    def fetch_label(self, label_name):
+        index = int(label_name.split('-')[1])
+        if self.labels[index] is None:
+            label = self._bmp_loader(join(self.root, self.LOCATION, str(index+1), 'gt'))
+            label = (label / 255.0).astype(np.uint8)    # To 0,1
+            self.labels[index] = label if self.phase == 'train' else self.cropper(label)
+        return self.labels[index]
+
+    @staticmethod
+    def _bmp_loader(bmp_path_wo_ext):
+        # Case insensitive .bmp loader
+        try:
+            return default_loader(bmp_path_wo_ext+'.bmp')
+        except FileNotFoundError:
+            return default_loader(bmp_path_wo_ext+'.BMP')
+
+        
\ No newline at end of file