def main():
args = get_args()
os.environ["CUDA_VISIBLE_DEVICES"] = args.gpus
SEED = 42
utils.set_global_seed(SEED)
utils.prepare_cudnn(deterministic=True)
num_classes = 14
#define datasets
train_dataset = ChestXrayDataSet(
data_dir=args.path_to_images,
image_list_file=args.train_list,
transform=transforms_train,
)
val_dataset = ChestXrayDataSet(
data_dir=args.path_to_images,
image_list_file=args.val_list,
transform=transforms_val,
)
loaders = {
'train':
DataLoader(train_dataset,
batch_size=args.batch_size,
shuffle=True,
num_workers=args.num_workers),
'valid':
DataLoader(val_dataset,
batch_size=2,
shuffle=False,
num_workers=args.num_workers)
}
logdir = args.log_dir #where model weights and logs are stored
#define model
model = DenseNet121(num_classes)
if len(args.gpus) > 1:
model = nn.DataParallel(model)
device = utils.get_device()
runner = SupervisedRunner(device=device)
optimizer = RAdam(model.parameters(), lr=args.lr, weight_decay=0.0003)
scheduler = optim.lr_scheduler.ReduceLROnPlateau(optimizer,
factor=0.25,
patience=2)
weights = torch.Tensor(
[10, 100, 30, 8, 40, 40, 330, 140, 35, 155, 110, 250, 155,
200]).to(device)
criterion = BCEWithLogitsLoss(pos_weight=weights)
class_names = [
'Atelectasis', 'Cardiomegaly', 'Effusion', 'Infiltration', 'Mass',
'Nodule', 'Pneumonia', 'Pneumothorax', 'Consolidation', 'Edema',
'Emphysema', 'Fibrosis', 'Pleural_Thickening', 'Hernia'
]
runner.train(
model=model,
logdir=logdir,
criterion=criterion,
optimizer=optimizer,
scheduler=scheduler,
loaders=loaders,
num_epochs=args.epochs,
# We can specify the callbacks list for the experiment;
# For this task, we will check AUC and accuracy
callbacks=[
AUCCallback(
input_key="targets",
output_key='logits',
prefix='auc',
class_names=class_names,
num_classes=num_classes,
activation='Sigmoid',
),
AccuracyCallback(
input_key="targets",
output_key="logits",
prefix="accuracy",
accuracy_args=[1],
num_classes=14,
threshold=0.5,
activation='Sigmoid',
),
],
main_metric='auc/_mean',
minimize_metric=False,
verbose=True,
)
def main():
cudnn.benchmark = True
model = DenseNet121(N_CLASSES).cuda()
model = torch.nn.DataParallel(model).cuda()
# data preprocess
normalize = transforms.Normalize([0.485, 0.456, 0.406],
[0.229, 0.224, 0.225])
train_transform = transforms.Compose([
transforms.Resize(256),
transforms.RandomCrop(224),
transforms.RandomHorizontalFlip(),
transforms.ToTensor(),
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
])
test_transform = transforms.Compose([
transforms.Resize(256),
# crop ten images from original
transforms.TenCrop(224),
transforms.Lambda(lambda crops: torch.stack(
[transforms.ToTensor()(crop) for crop in crops])),
transforms.Lambda(
lambda crops: torch.stack([normalize(crop) for crop in crops]))
])
# load data
if params["aug_dataset"]:
train_dataset = ChestXrayDataSetWithAugmentation(
data_dir=DATA_DIR,
image_list_file=TRAIN_IMAGE_LIST,
transform=train_transform)
else:
train_dataset = ChestXrayDataSet(data_dir=DATA_DIR,
image_list_file=TRAIN_IMAGE_LIST,
transform=train_transform)
train_loader = DataLoader(dataset=train_dataset,
batch_size=params["train_batch_size"],
shuffle=True,
num_workers=8,
pin_memory=True)
train_evaluation_dataset = ChestXrayDataSet(
data_dir=DATA_DIR,
image_list_file=TRAIN_IMAGE_LIST,
transform=test_transform)
train_evaluation_loader = DataLoader(dataset=train_evaluation_dataset,
batch_size=params["test_batch_size"],
shuffle=False,
num_workers=8,
pin_memory=True)
dev_dataset = ChestXrayDataSet(data_dir=DATA_DIR,
image_list_file=DEV_IMAGE_LIST,
transform=test_transform)
dev_loader = DataLoader(dataset=dev_dataset,
batch_size=params["test_batch_size"],
shuffle=False,
num_workers=8,
pin_memory=True)
test_dataset = ChestXrayDataSet(data_dir=DATA_DIR,
image_list_file=TEST_IMAGE_LIST,
transform=test_transform)
test_loader = DataLoader(dataset=test_dataset,
batch_size=params["test_batch_size"],
shuffle=False,
num_workers=8,
pin_memory=True)
optimizer = torch.optim.Adam(model.parameters(),
lr=params["lr"],
weight_decay=params["beta"])
def train(epoch):
print("start training epoch %d" % (epoch))
start_time = time()
local_step = 0
running_loss = 0
running_loss_list = []
model.train()
if not params["aug_dataset"]:
pos_weight = torch.tensor(train_dataset.pos_weight,
dtype=torch.float32).cuda()
neg_weight = torch.tensor(train_dataset.neg_weight,
dtype=torch.float32).cuda()
class_weight = torch.tensor(train_dataset.class_weight,
dtype=torch.float32).cuda()
for i, (inp, target) in enumerate(train_loader):
inp = inp.cuda()
target = target.cuda()
optimizer.zero_grad()
output = model(inp)
if params["loss_function"] == "unweighted":
local_loss = F.binary_cross_entropy(output, target)
elif params["loss_function"] == "weighted1":
local_loss = weighted_binary_cross_entropy(
output, target, pos_weight, neg_weight)
elif params["loss_function"] == "weighted2":
local_loss = weighted_binary_cross_entropy2(
output, target, class_weight)
else:
assert False
running_loss += local_loss.item()
local_loss.backward()
optimizer.step()
if (i + 1) % PRINT_FREQ == 0:
running_loss /= PRINT_FREQ
print("epoch %d, batch %d/%d, loss: %.5f" %
(epoch, i + 1, len(train_loader), running_loss))
running_loss_list.append(running_loss)
running_loss = 0
print("end training epoch %d, time elapsed: %.2fmin" %
(epoch, (time() - start_time) / 60))
return dict(running_loss_list=running_loss_list)
def evaluate(epoch, dataset_loader, pytorch_dataset, dataset_name):
print("start evaluating epoch %d on %s" % (epoch, dataset_name))
gt = torch.tensor([], dtype=torch.float32, device="cuda")
pred = torch.tensor([], dtype=torch.float32, device="cuda")
loss = 0.
model.eval()
if not params["aug_dataset"]:
pos_weight = torch.tensor(pytorch_dataset.pos_weight,
dtype=torch.float32).cuda()
neg_weight = torch.tensor(pytorch_dataset.neg_weight,
dtype=torch.float32).cuda()
class_weight = torch.tensor(train_dataset.class_weight,
dtype=torch.float32).cuda()
with torch.no_grad():
for i, (inp, target) in enumerate(dataset_loader):
target = target.cuda()
gt = torch.cat((gt, target), 0)
bs, n_crops, c, h, w = inp.size()
inp_reshaped = inp.view(-1, c, h, w).cuda()
output = model(inp_reshaped)
output_mean = output.view(bs, n_crops, -1).mean(1)
pred = torch.cat((pred, output_mean), 0)
if params["loss_function"] == "unweighted":
local_loss = F.binary_cross_entropy(output_mean, target)
elif params["loss_function"] == "weighted1":
local_loss = weighted_binary_cross_entropy(
output_mean, target, pos_weight, neg_weight)
elif params["loss_function"] == "weighted2":
local_loss = weighted_binary_cross_entropy2(
output_mean, target, class_weight)
else:
assert False
loss += local_loss * len(target) / len(pytorch_dataset)
AUROCs = compute_AUCs(gt, pred, N_CLASSES)
AUROC_avg = np.array(AUROCs).mean()
print("epoch %d, %s, loss: %.5f, avg_AUC: %.5f" %
(epoch, dataset_name, loss, AUROC_avg))
print("epoch %d, %s, individual class AUC" % (epoch, dataset_name))
for i in range(N_CLASSES):
print('\tthe AUROC of %s is %.5f' % (CLASS_NAMES[i], AUROCs[i]))
return dict(auroc=dict(zip(CLASS_NAMES, AUROCs)),
auroc_avg=AUROC_avg,
loss=loss.item())
def init_history():
return dict(epoch=0,
train_eval_vals_list=[],
dev_eval_vals_list=[],
best_dev_eval_vals=dict(auroc_avg=-np.inf, loss=np.inf),
best_dev_eval_vals_epoch=-1)
def update_history(history, epoch, train_eval_vals, dev_eval_vals):
history["epoch"] = epoch
history["train_eval_vals_list"].append(train_eval_vals)
history["dev_eval_vals_list"].append(dev_eval_vals)
if dev_eval_vals["auroc_avg"] > history["best_dev_eval_vals"][
"auroc_avg"]:
history["best_dev_eval_vals"] = dev_eval_vals
history["best_dev_eval_vals_epoch"] = epoch
if epoch >= 1:
print("saving model...")
state_dict = model.state_dict()
torch.save(state_dict, params["best_model_file_path"])
if epoch >= 1:
state_dict = model.state_dict()
torch.save(state_dict, params["model_file_path"] % (epoch))
with open(params["history_file_path"], 'wb') as f:
pickle.dump(history, f)
def train_initialization():
if not os.path.exists(params["base_dir"]):
os.mkdir(params["base_dir"])
with open(params["params_file_path"], 'w') as f:
yaml.dump(params, f, default_flow_style=False)
if os.path.exists(params["history_file_path"]):
with open(params["history_file_path"], 'rb') as f:
old_history = pickle.load(f)
last_epoch = old_history["epoch"]
if last_epoch > params["epochs"]:
print("training completed")
exit(0)
model_file = params["model_file_path"] % (last_epoch)
if os.path.exists(model_file):
model.load_state_dict(torch.load(model_file))
print("training resumed from epoch %d" % last_epoch)
return old_history, last_epoch
return init_history(), 0
history, last_epoch = train_initialization()
for epoch in range(last_epoch + 1, params["epochs"] + 1):
train_eval_vals = train(epoch)
train_eval_vals2 = evaluate(epoch, train_evaluation_loader,
train_evaluation_dataset, "train set")
dev_eval_vals = evaluate(epoch, dev_loader, dev_dataset, "dev set")
update_history(history, epoch, train_eval_vals, dev_eval_vals)
print("training completed")
def main():
N_CLASSES = 14
CLASS_NAMES = ['Atelectasis',
'Cardiomegaly',
'Effusion',
'Infiltration',
'Mass',
'Nodule',
'Pneumonia',
'Pneumothorax',
'Consolidation',
'Edema',
'Emphysema',
'Fibrosis',
'Pleural_Thickening',
'Hernia']
# initialize model
device = utils.get_device()
model = DenseNet121(N_CLASSES).to(device)
checkpoint = torch.load(args.checkpoint)
model.load_state_dict(checkpoint['model_state_dict'])
# initialize test loader
test_dataset = ChestXrayDataSet(data_dir=args.path_to_images,
image_list_file=args.test_list,
transform=transforms_test)
test_loader = DataLoader(dataset=test_dataset, batch_size=args.batch_size,
shuffle=False, num_workers=args.num_workers, pin_memory=True)
# initialize the ground truth and output tensor
gt = torch.FloatTensor()
gt = gt.cuda()
pred = torch.FloatTensor()
pred = pred.cuda()
# switch to evaluate mode
model.eval()
with torch.no_grad():
for i, (inp, target) in enumerate(test_loader):
target = target.cuda()
gt = torch.cat((gt, target), 0)
bs, c, h, w = inp.size()
input_var = torch.autograd.Variable(inp.view(-1, c, h, w).cuda())
output = model(input_var)
output_mean = output.view(bs, -1)
pred = torch.cat((pred, output_mean.data), 0)
gt_np = gt.cpu().numpy()
pred_np = sigmoid(pred.cpu().numpy())
Y_t = [] #labels for each anomaly
for i in range(N_CLASSES):
Y_t.append([])
for x in gt_np:
Y_t[i].append(x[i])
Y_pred = [] #preds for each anomaly
for j in range(N_CLASSES):
Y_pred.append([])
for y in pred_np:
Y_pred[j].append(y[j])
AUCs = [] # AUCs for each
for i in range(N_CLASSES):
auc = roc_auc_score(Y_t[i], Y_pred[i])
AUCs.append(auc)
matrices=[] #for each
for i in range(14):
matrix = confusion_matrix(Y_t[i], np.asarray(Y_pred[i])>0.6)
matrices.append(matrix)
class_names = ['no disease', 'disease']
fig = plt.figure(figsize = (20,20))
for i in range(14):
plt.subplot(4,4,i+1)
df_cm = pd.DataFrame(
matrices[i], index=class_names, columns=class_names)
heatmap = sns.heatmap(df_cm, annot=True, fmt="d").set_title(CLASS_NAMES[i])
plt.ylabel('True label')
plt.xlabel('Predicted label')
plt.show()
fig.savefig(os.path.join(args.test_outdir,'confusion_matrix.pdf'))
fig, axes2d = plt.subplots(nrows=2, ncols=7,
sharex=True, sharey=True,figsize = (12, 4))
for i, row in enumerate(axes2d):
for j, cell in enumerate(row):
if i==0:
x=i+j
else:
x=13-i*j
fpr, tpr, threshold = roc_curve(Y_t[x], Y_pred[x])
roc_auc = auc(fpr, tpr)
cell.plot(fpr, tpr, 'b', label = 'AUC = %0.2f' % roc_auc)
cell.legend(loc = 'lower right', handlelength=0,handletextpad=0,frameon=False, prop={'size': 8})
cell.plot([0, 1], [0, 1],'r--')
plt.xlim([0, 1])
plt.ylim([0, 1])
cell.set_title(CLASS_NAMES[x],fontsize=10)
if i == len(axes2d) - 1:
cell.set_xlabel('False positive rate')
if j == 0:
cell.set_ylabel('True negative rate')
fig.tight_layout(pad=1.0)
plt.show()
fig.savefig(os.path.join(args.test_outdir,'roc_auc.pdf'))
DATA_DIR = '/home/lrh/dataset/ChestXray-NIHCC/images_v1_small'
testTXTFile = '/home/lrh/git/CheXNet/ChestX-ray14/labels/test.txt'
trainTXTFile = '/home/lrh/git/CheXNet/ChestX-ray14/labels/train.txt'
normalize = transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])
train_transform = transforms.Compose([
transforms.Resize(256),
transforms.CenterCrop(224),
transforms.ToTensor(), normalize
])
test_transform = transforms.Compose([
transforms.Resize(256),
transforms.CenterCrop(224),
transforms.ToTensor(), normalize
])
train_dataset = ChestXrayDataSet(data_dir=DATA_DIR,
image_list_file=trainTXTFile,
transform=train_transform)
train_loader = torch.utils.data.DataLoader(dataset=train_dataset,
batch_size=16,
shuffle=True,
num_workers=8,
pin_memory=True)
test_dataset = ChestXrayDataSet(data_dir=DATA_DIR,
image_list_file=testTXTFile,
transform=test_transform)
test_loader = torch.utils.data.DataLoader(dataset=test_dataset,
batch_size=16,
shuffle=False,
num_workers=8,
pin_memory=True)
print('\ndone')
def train(self,
TRAIN_IMAGE_LIST,
VAL_IMAGE_LIST,
NUM_EPOCHS=10,
LR=0.001,
BATCH_SIZE=64,
start_epoch=0,
logging=True,
save_path=None,
freeze_feature_layers=True):
"""
Train the CovidAID
"""
normalize = transforms.Normalize([0.485, 0.456, 0.406],
[0.229, 0.224, 0.225])
train_dataset = ChestXrayDataSet(
image_list_file=TRAIN_IMAGE_LIST,
transform=transforms.Compose([
transforms.Resize(256),
transforms.TenCrop(224),
transforms.Lambda(lambda crops: torch.stack(
[transforms.ToTensor()(crop) for crop in crops])),
transforms.Lambda(lambda crops: torch.stack(
[normalize(crop) for crop in crops]))
]),
combine_pneumonia=self.combine_pneumonia)
if self.distributed:
sampler = DistributedSampler(train_dataset)
train_loader = DataLoader(dataset=train_dataset,
batch_size=BATCH_SIZE,
shuffle=False,
num_workers=8,
pin_memory=True,
sampler=sampler)
else:
train_loader = DataLoader(dataset=train_dataset,
batch_size=BATCH_SIZE,
shuffle=True,
num_workers=8,
pin_memory=True)
val_dataset = ChestXrayDataSet(
image_list_file=VAL_IMAGE_LIST,
transform=transforms.Compose([
transforms.Resize(256),
transforms.TenCrop(224),
transforms.Lambda(lambda crops: torch.stack(
[transforms.ToTensor()(crop) for crop in crops])),
transforms.Lambda(lambda crops: torch.stack(
[normalize(crop) for crop in crops]))
]),
combine_pneumonia=self.combine_pneumonia)
if self.distributed:
sampler = DistributedSampler(val_dataset)
val_loader = DataLoader(dataset=val_dataset,
batch_size=BATCH_SIZE,
shuffle=False,
num_workers=8,
pin_memory=True,
sampler=sampler)
else:
val_loader = DataLoader(dataset=val_dataset,
batch_size=BATCH_SIZE,
shuffle=True,
num_workers=8,
pin_memory=True)
# Freeze heads and create optimizer
if freeze_feature_layers:
print("Freezing feature layers")
for param in self.net.densenet121.features.parameters():
param.requires_grad = False
# optimizer = optim.SGD(filter(lambda p: p.requires_grad, self.net.parameters()),
# lr=LR, momentum=0.9)
optimizer = optim.Adam(filter(lambda p: p.requires_grad,
self.net.parameters()),
lr=LR)
for epoch in range(start_epoch, NUM_EPOCHS):
# switch to train mode
self.net.train()
tot_loss = 0.0
for i, (inputs,
target) in tqdm(enumerate(train_loader),
total=len(train_dataset) / BATCH_SIZE):
# inputs = inputs.to(self.device)
# target = target.to(self.device)
inputs = inputs.cuda()
target = target.cuda()
# Shape of input == [BATCH_SIZE, NUM_CROPS=19, CHANNELS=3, HEIGHT=224, WIDTH=244]
bs, n_crops, c, h, w = inputs.size()
inputs = inputs.view(-1, c, h, w)
inputs = torch.autograd.Variable(inputs.view(-1, c, h, w))
target = torch.autograd.Variable(target)
preds = self.net(inputs).view(bs, n_crops, -1).mean(dim=1)
# loss = torch.sum(torch.abs(preds - target) ** 2)
loss = train_dataset.loss(preds, target)
# exit()
tot_loss += float(loss.data)
optimizer.zero_grad()
loss.backward()
optimizer.step()
tot_loss /= len(train_dataset)
# Clear cache
torch.cuda.empty_cache()
# Running on validation set
self.net.eval()
val_loss = 0.0
for i, (inputs,
target) in tqdm(enumerate(val_loader),
total=len(val_dataset) / BATCH_SIZE):
# inputs = inputs.to(self.device)
# target = target.to(self.device)
inputs = inputs.cuda()
target = target.cuda()
# Shape of input == [BATCH_SIZE, NUM_CROPS=19, CHANNELS=3, HEIGHT=224, WIDTH=244]
bs, n_crops, c, h, w = inputs.size()
inputs = inputs.view(-1, c, h, w)
inputs = torch.autograd.Variable(inputs.view(-1, c, h, w),
volatile=True)
target = torch.autograd.Variable(target, volatile=True)
preds = self.net(inputs).view(bs, n_crops, -1).mean(1)
# loss = torch.sum(torch.abs(preds - target) ** 2)
loss = val_dataset.loss(preds, target)
val_loss += float(loss.data)
val_loss /= len(val_dataset)
# Clear cache
torch.cuda.empty_cache()
# logging statistics
timestamp = str(datetime.datetime.now()).split('.')[0]
log = json.dumps({
'timestamp': timestamp,
'epoch': epoch + 1,
'train_loss': float('%.5f' % tot_loss),
'val_loss': float('%.5f' % val_loss),
'lr': float('%.6f' % LR)
})
if logging:
print(log)
log_file = os.path.join(save_path, 'train.log')
if log_file is not None:
with open(log_file, 'a') as f:
f.write("{}\n".format(log))
model_path = os.path.join(save_path, 'epoch_%d.pth' % (epoch + 1))
self.save_model(model_path)
print('Finished Training')