solver.py 12.6 KB
Newer Older
1
2
3
4
"""Brain Mapper U-Net Solver

Description:

5
    This folder contains the Pytorch implementation of the core U-net solver, used for training the network.
6

7
Usage:
8

9
10
11
    To use this module, import it and instantiate is as you wish:

        from solver import Solver
12
13
14
15
16
"""

import os
import numpy as np
import torch
17
18
import glob

19
from datetime import datetime
20
21
from utils.losses import MSELoss
from utils.data_utils import create_folder
22
from utils.data_logging_utils import LogWriter
23
from torch.optim import lr_scheduler
24
25
26
27

checkpoint_directory = 'checkpoints'
checkpoint_extension = 'path.tar'

28

29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
class Solver():
    """Solver class for the BrainMapper U-net.

    This class contains the pytorch implementation of the U-net solver required for the BrainMapper project.

    Args:
        model (class): BrainMapper model class
        experiment_name (str): Name of the experiment
        device (int/str): Device type used for training (int - GPU id, str- CPU)
        number_of_classes (int): Number of classes
        optimizer (class): Pytorch class of desired optimizer
        optimizer_arguments (dict): Dictionary of arguments to be optimized
        loss_function (func): Function describing the desired loss function
        model_name (str): Name of the model
        labels (arr): Vector/Array of labels (if applicable)
        number_epochs (int): Number of training epochs
        loss_log_period (int): Period for writing loss value
        learning_rate_scheduler_step_size (int): Period of learning rate decay
        learning_rate_scheduler_gamma (int): Multiplicative factor of learning rate decay
        use_last_checkpoint (bool): Flag for loading the previous checkpoint
        experiment_directory (str): Experiment output directory name
        logs_directory (str): Directory for outputing training logs

    Returns:
        trained model(?) - working on this!

    """
56

57
    def __init__(self,
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
                 model,
                 device,
                 number_of_classes,
                 experiment_name,
                 optimizer=torch.optim.Adam,
                 optimizer_arguments={},
                 loss_function=MSELoss(),
                 model_name='BrainMapper',
                 labels=None,
                 number_epochs=10,
                 loss_log_period=5,
                 learning_rate_scheduler_step_size=5,
                 learning_rate_scheduler_gamma=0.5,
                 use_last_checkpoint=True,
                 experiment_directory='experiments',
                 logs_directory='logs'
                 ):
75
76
77

        self.model = model
        self.device = device
78
79
80
81
82
83
84
85
86
87
88
        self.optimizer = optimizer(model.parameters(), **optimizer_arguments)

        if torch.cuda.is_available():
            self.loss_function = loss_function.cuda(device)
        else:
            self.loss_function = loss_function

        self.model_name = model_name
        self.labels = labels
        self.number_epochs = number_epochs
        self.loss_log_period = loss_log_period
89

90
91
        # We use a learning rate scheduler, that decays the LR of each paramter group by gamma every step_size epoch.
        self.learning_rate_scheduler = lr_scheduler.StepLR(self.optimizer,
92
93
94
                                                           step_size=learning_rate_scheduler_step_size,
                                                           gamma=learning_rate_scheduler_gamma)

95
96
        self.use_last_checkpoint = use_last_checkpoint

97
        experiment_directory_path = os.path.join(
98
            experiment_directory, experiment_name)
99
        self.experiment_directory_path = experiment_directory_path
100

101
        create_folder(experiment_directory)
102
        create_folder(experiment_directory_path)
103
        create_folder(os.path.join(
104
            experiment_directory_path, checkpoint_directory))
105
106
107
108

        self.start_epoch = 1
        self.start_iteration = 1
        self.best_mean_score = 0
109
        self.best_mean_score_epoch = 0
110
111
112

        if use_last_checkpoint:
            self.load_checkpoint()
113

114
115
116
117
118
        self.LogWriter = LogWriter(number_of_classes=number_of_classes,
                                   logs_directory=logs_directory,
                                   experiment_name=experiment_name,
                                   use_last_checkpoint=use_last_checkpoint,
                                   labels=labels)
119

120
121
122
123
124
125
126
127
128
129
    def train(self, train_loader, test_loader):
        """Training Function

        This function trains a given model using the provided training data.

        Args:
            train_loader (class): Combined dataset and sampler, providing an iterable over the training dataset (torch.utils.data.DataLoader)
            test_loader (class):  Combined dataset and sampler, providing an iterable over the testing dataset (torch.utils.data.DataLoader)

        Returns:
130
            trained model
131
132
133
134
135
136
        """

        model, optimizer, learning_rate_scheduler = self.model, self.optimizer, self.learning_rate_scheduler
        dataloaders = {'train': train_loader, 'test': test_loader}

        if torch.cuda.is_available():
137
138
            torch.cuda.empty_cache()  # clear memory
            model.cuda(self.device)  # Moving the model to GPU
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168

        print('****************************************************************')
        print('TRAINING IS STARTING!')
        print('=====================')
        print('Model Name: {}'.format(self.model_name))
        print('Device Type: {}'.format(torch.cuda.get_device_name(self.device)))
        start_time = datetime.now()
        print('Started At: {}'.format(start_time))
        print('----------------------------------------')

        iteration = self.start_iteration

        for epoch in range(self.start_epoch, self.number_epochs+1):
            print("Epoch {}/{}".format(epoch, self.number_epochs))

            for phase in ['train', 'test']:
                print('-> Phase: {}'.format(phase))

                losses = []
                outputs = []
                y_values = []

                if phase == 'train':
                    model.train()
                    learning_rate_scheduler.step()
                else:
                    model.eval()

                for batch_index, sampled_batch in enumerate(dataloaders[phase]):
                    X = sampled_batch[0].type(torch.FloatTensor)
169
                    y = sampled_batch[1].type(torch.LongTensor)
170

171
172
173
174
                    # We add an extra dimension (~ number of channels) for the 3D convolutions.
                    X = torch.unsqueeze(X, dim= 1)
                    y = torch.unsqueeze(y, dim= 1)

175
                    if model.is_cuda():
176
177
                        X = X.cuda(self.device, non_blocking=True)
                        y = y.cuda(self.device, non_blocking=True)
178
179
180

                    y_hat = model(X)   # Forward pass

181
                    loss = self.loss_function(y_hat, y)  # Loss computation
182
183

                    if phase == 'train':
184
185
                        optimizer.zero_grad()  # Zero the parameter gradients
                        loss.backward()  # Backward propagation
186
187
188
189
                        optimizer.step()

                        if batch_index % self.loss_log_period == 0:

190
191
                            self.LogWriter.loss_per_iteration(
                                loss.item(), batch_index, iteration)
192

193
194
195
                        iteration += 1

                    losses.append(loss.item())
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
                    outputs.append(torch.max(y_hat, dim=1)[1].cpu())
                    y_values.append(y.cpu())

                    # Clear the memory

                    del X, y, y_hat, loss
                    torch.cuda.empty_cache()

                    if phase == 'test':
                        if batch_index != len(dataloaders[phase]) - 1:
                            print("#", end='', flush=True)
                        else:
                            print("100%", flush=True)

                with torch.no_grad():
211
212
                    output_array, y_array = torch.cat(
                        outputs), torch.cat(y_values)
213

214
215
                    self.LogWriter.loss_per_epoch(losses, phase, epoch)

216
217
                    dice_score_mean = self.LogWriter.dice_score_per_epoch(
                        phase, output_array, y_array, epoch)
218
                    if phase == 'test':
219
220
221
                        if dice_score_mean > self.best_mean_score:
                            self.best_mean_score = dice_score_mean
                            self.best_mean_score_epoch = epoch
222

223
224
225
226
227
228
                    index = np.random.choice(
                        len(dataloaders[phase].dataset.X), size=3, replace=False)
                    self.LogWriter.sample_image_per_epoch(prediction=model.predict(dataloaders[phase].dataset.X[index], self.device),
                                                          ground_truth=dataloaders[phase].dataset.y[index],
                                                          phase=phase,
                                                          epoch=epoch)
229
230

            print("Epoch {}/{} DONE!".format(epoch, self.number_epochs))
231

232
233
234
235
236
237
238
            self.save_checkpoint(state={'epoch': epoch + 1,
                                        'start_iteration': iteration + 1,
                                        'arch': self.model_name,
                                        'state_dict': model.state_dict(),
                                        'optimizer': optimizer.state_dict(),
                                        'scheduler': learning_rate_scheduler.state_dict()
                                        },
239
240
241
                                 filename=os.path.join(self.experiment_directory_path, checkpoint_directory,
                                                       'checkpoint_epoch_' + str(epoch) + '.' + checkpoint_extension)
                                 )
242

243
        self.LogWriter.close()
244
245
246
247
248
249
250
251
252

        print('----------------------------------------')
        print('TRAINING IS COMPLETE!')
        print('=====================')
        end_time = datetime.now()
        print('Completed At: {}'.format(end_time))
        print('Training Duration: {}'.format(end_time - start_time))
        print('****************************************************************')

253
254
255
256
257
258
259
260
261
262
    def save_checkpoint(self, state, filename):
        """General Checkpoint Save

        This function saves a general checkpoint for inference and/or resuming training

        Args:
            state (dict): Dictionary of all the relevant model components
        """

        torch.save(state, filename)
263

264
    def load_checkpoint(self, epoch=None):
265
266
267
268
269
270
271
        """General Checkpoint Loader

        This function loads a previous checkpoint for inference and/or resuming training

        Args:
            epoch (int): Current epoch value
        """
272

273
        if epoch is None:
274
275
            checkpoint_file_path = os.path.join(
                self.experiment_directory_path, checkpoint_directory, 'checkpoint_epoch_' + str(epoch) + '.' + checkpoint_extension)
276
277
            self._checkpoint_reader(checkpoint_file_path)
        else:
278
279
            universal_path = os.path.join(
                self.experiment_directory_path, checkpoint_directory, '*.' + checkpoint_extension)
280
281
282
283
284
            files_in_universal_path = glob.glob(universal_path)

            # We will sort through all the files in path to see which one is most recent

            if len(files_in_universal_path) > 0:
285
286
                checkpoint_file_path = max(
                    files_in_universal_path, key=os.path.getatime)
287
288
289
                self._checkpoint_reader(checkpoint_file_path)

            else:
290
291
                self.LogWriter.log("No Checkpoint found at {}".format(
                    os.path.join(self.experiment_directory_path, checkpoint_directory)))
292

293
294
295
296
297
298
299
300
301
    def _checkpoint_reader(self, checkpoint_file_path):
        """Checkpoint Reader

        This private function reads a checkpoint file and then loads the relevant variables

        Args:
            checkpoint_file_path (str): path to checkpoint file
        """

302
303
        self.LogWriter.log(
            "Loading Checkpoint {}".format(checkpoint_file_path))
304
305
306
307

        checkpoint = torch.load(checkpoint_file_path)
        self.start_epoch = checkpoint['epoch']
        self.start_iteration = checkpoint['start_iteration']
308
        # We are not loading the model_name as we might want to pre-train a model and then use it.
309
310
        self.model.load_state_dict = checkpoint['state_dict']
        self.optimizer.load_state_dict = checkpoint['optimizer']
311
        self.learning_rate_scheduler.load_state_dict = checkpoint['scheduler']
312
313

        for state in self.optimizer.state.values():
314
            for key, value in state.items():
315
316
317
                if torch.is_tensor(value):
                    state[key] = value.to(self.device)

318
319
        self.LogWriter.log(
            "Checkpoint Loaded {} - epoch {}".format(checkpoint_file_path, checkpoint['epoch']))