net,
args.input_size,
batch_size=args.batch_size,
crf_iterations=args.crf_iterations,
mu=mu,
std=std)
"""
Validate the segmentation
"""
print('[%s] Validating segmentation' % (datetime.datetime.now()))
test_data_labels = read_tif(args.data_labels, dtype='uint8')
test_data_labels = normalize(test_data_labels, 0, 255)
j = jaccard(segmentation, test_data_labels)
d = dice(segmentation, test_data_labels)
a, p, r, f = accuracy_metrics(segmentation, test_data_labels)
print('[%s] RESULTS:' % (datetime.datetime.now()))
print('[%s] Jaccard: %f' % (datetime.datetime.now(), j))
print('[%s] Dice: %f' % (datetime.datetime.now(), d))
print('[%s] Accuracy: %f' % (datetime.datetime.now(), a))
print('[%s] Precision: %f' % (datetime.datetime.now(), p))
print('[%s] Recall: %f' % (datetime.datetime.now(), r))
print('[%s] F-score: %f' % (datetime.datetime.now(), f))
"""
Write out the results
"""
if args.write_dir is not None:
print('[%s] Writing the output' % (datetime.datetime.now()))
imwrite3D(segmentation, args.write_dir, rescale=True)
print('[%s] Finished!' % (datetime.datetime.now()))
Setup optimization for unsupervised training
"""
print('[%s] Setting up optimization for unsupervised training' % (datetime.datetime.now()))
net = UNetAE(feature_maps=args.fm, levels=args.levels, group_norm=(args.group_norm == 1), lambda_dom=args.lambda_dom, s=args.input_size[0])
"""
Train the network unsupervised
"""
print('[%s] Training network unsupervised' % (datetime.datetime.now()))
net.train_net(src_train_loader=src_train_loader, src_test_loader=src_test_loader,
tar_train_loader=tar_train_loader, tar_test_loader=tar_test_loader,
loss_fn=loss_fn_rec, lr=args.lr, step_size=args.step_size, gamma=args.gamma,
epochs=args.epochs, test_freq=args.test_freq, print_stats=args.print_stats,
log_dir=args.log_dir)
"""
Write out the results
"""
if args.write_dir is not None:
print('[%s] Running the trained network on the source dataset' % (datetime.datetime.now()))
src_data = src_test.data
transform_last_checkpoint = transform(src_data, net, args.input_size, batch_size=args.test_batch_size)
print('[%s] Writing the output' % (datetime.datetime.now()))
imwrite3D(transform_last_checkpoint, os.path.join(args.write_dir, 'src_transform'), rescale=True)
tar_data = tar_test.data
print('[%s] Running the trained network on the target dataset' % (datetime.datetime.now()))
transform_last_checkpoint = transform(tar_data, net, args.input_size, batch_size=args.test_batch_size)
print('[%s] Writing the output' % (datetime.datetime.now()))
imwrite3D(transform_last_checkpoint, os.path.join(args.write_dir, 'tar_transform'), rescale=True)
print('[%s] Finished!' % (datetime.datetime.now()))
print('[%s] F-score: %f' % (datetime.datetime.now(), f))
net = torch.load(os.path.join(args.log_dir, 'best_checkpoint.pytorch'))
segmentation_best_checkpoint = segment_pixels(test_data,
net,
args.input_size,
batch_size=args.test_batch_size)
j = jaccard(segmentation_best_checkpoint, test_labels)
d = dice(segmentation_best_checkpoint, test_labels)
a, p, r, f = accuracy_metrics(segmentation_best_checkpoint, test_labels)
print('[%s] RESULTS:' % (datetime.datetime.now()))
print('[%s] Jaccard: %f' % (datetime.datetime.now(), j))
print('[%s] Dice: %f' % (datetime.datetime.now(), d))
print('[%s] Accuracy: %f' % (datetime.datetime.now(), a))
print('[%s] Precision: %f' % (datetime.datetime.now(), p))
print('[%s] Recall: %f' % (datetime.datetime.now(), r))
print('[%s] F-score: %f' % (datetime.datetime.now(), f))
print('[%s] Network performance (best checkpoint): Jaccard=%f - Dice=%f' %
(datetime.datetime.now(), j, d))
"""
Write out the results
"""
if args.write_dir is not None:
print('[%s] Writing the output' % (datetime.datetime.now()))
imwrite3D(segmentation_last_checkpoint,
os.path.join(args.write_dir, 'segmentation_last_checkpoint'),
rescale=True)
imwrite3D(segmentation_best_checkpoint,
os.path.join(args.write_dir, 'segmentation_best_checkpoint'),
rescale=True)
print('[%s] Finished!' % (datetime.datetime.now()))
net = torch.load(os.path.join(log_dir, 'best_checkpoint.pytorch'))
segmentation_best_checkpoint = segment(test_data,
net,
input_size,
batch_size=args.test_batch_size)
j = jaccard(segmentation_best_checkpoint, test_labels)
d = dice(segmentation_best_checkpoint, test_labels)
a, p, r, f = accuracy_metrics(segmentation_best_checkpoint,
test_labels)
print('[%s] Results best checkpoint:' % (datetime.datetime.now()))
print('[%s] Jaccard: %f' % (datetime.datetime.now(), j))
print('[%s] Dice: %f' % (datetime.datetime.now(), d))
print('[%s] Accuracy: %f' % (datetime.datetime.now(), a))
print('[%s] Precision: %f' % (datetime.datetime.now(), p))
print('[%s] Recall: %f' % (datetime.datetime.now(), r))
print('[%s] F-score: %f' % (datetime.datetime.now(), f))
"""
Write out the results
"""
if args.write_dir is not None:
print('[%s] Writing the output' % (datetime.datetime.now()))
imwrite3D(segmentation_last_checkpoint,
os.path.join(write_dir, 'segmentation_last_checkpoint',
str(o)),
rescale=True)
imwrite3D(segmentation_best_checkpoint,
os.path.join(write_dir, 'segmentation_best_checkpoint',
str(o)),
rescale=True)
print('[%s] Finished!' % (datetime.datetime.now()))