from __future__ import print_function
import os,sys,inspect
currentdir = os.path.dirname(os.path.abspath(inspect.getfile(inspect.currentframe())))
parentdir = os.path.dirname(os.path.dirname(currentdir))
sys.path.insert(0,parentdir)
from utils.args import args
import setup.categories.ae_setup as AESetup
from models.autoencoders import *
from datasets.NIH_Chest import NIHChestBinaryTrainSplit
if __name__ == "__main__":
dataset = NIHChestBinaryTrainSplit(root_path=os.path.join(args.root_path, "NIHCC"), binary=True, expand_channels=False, downsample=64)
model = Generic_VAE(dims=(1, 64, 64), max_channels=512, depth=12, n_hidden=512)
#model = ALILikeVAE(dims=(1, 64, 64))
AESetup.train_variational_autoencoder(args, model=model, dataset=dataset.get_D1_train(), BCE_Loss=False)
sys.path.insert(0, parentdir)
import models as Models
import global_vars as Global
from utils.args import args
import categories.classifier_setup as CLSetup
from models.classifiers import NIHDenseBinary, NIHChestVGG
from datasets.NIH_Chest import NIHChestBinaryTrainSplit
if __name__ == "__main__":
dataset = NIHChestBinaryTrainSplit(root_path=os.path.join(
args.root_path, "NIHCC"),
binary=True)
model = NIHChestVGG()
CLSetup.train_classifier(args, model=model, dataset=dataset.get_D1_train())
# task_list = [
# # The list of models, The function that does the training, Can I skip-test?, suffix of the operation.
# # The procedures that can be skip-test are the ones that we can determine
# # whether we have done them before without instantiating the network architecture or dataset.
# # saves quite a lot of time when possible.
# (Global.dataset_reference_classifiers, CLSetup.train_classifier, True, ['base0']),
# (Global.dataset_reference_classifiers, KLogisticSetup.train_classifier, True, ['KLogistic']),
# (Global.dataset_reference_classifiers, DeepEnsembleSetup.train_classifier, True, ['DE.%d'%i for i in range(5)]),
# (Global.dataset_reference_autoencoders, AESetup.train_BCE_AE, False, []),
# (Global.dataset_reference_autoencoders, AESetup.train_MSE_AE, False, []),
# (Global.dataset_reference_vaes, AESetup.train_variational_autoencoder, False, []),
# (Global.dataset_reference_pcnns, PCNNSetup.train_pixelcnn, False, []),
# ]
#
from __future__ import print_function
import os, sys, inspect
currentdir = os.path.dirname(
os.path.abspath(inspect.getfile(inspect.currentframe())))
parentdir = os.path.dirname(os.path.dirname(currentdir))
sys.path.insert(0, parentdir)
from utils.args import args
import setup.categories.ae_setup as AESetup
from models.autoencoders import Generic_VAE, Generic_AE, Residual_AE
from datasets.NIH_Chest import NIHChestBinaryTrainSplit
if __name__ == "__main__":
dataset = NIHChestBinaryTrainSplit(root_path=os.path.join(
args.root_path, "NIHCC"),
binary=True,
expand_channels=False,
downsample=64)
model = Residual_AE(dims=(1, 64, 64))
AESetup.train_autoencoder(args,
model=model,
dataset=dataset.get_D1_train(),
BCE_Loss=False)
if not args.load or not os.path.exists(
os.path.join(
args.experiment_path, "all_embs_UC3_ppd_%d_d1_%s.npy" %
(args.points_per_d2, args.dataset))):
assert args.dataset in ['NIHCC', 'PADChest']
if args.dataset.lower() == 'nihcc':
D164 = NIHChestBinaryTrainSplit(root_path=os.path.join(
args.root_path, 'NIHCC'),
downsample=64)
elif args.dataset.lower() == "padchest":
D164 = PADChestBinaryTrainSplit(root_path=os.path.join(
args.root_path, "PADChest"),
binary=True,
downsample=64)
D1 = D164.get_D1_train()
emb = args.embedding_function.lower()
assert emb in ["vae", "ae", "ali"]
dummy_args = EasyDict()
dummy_args.exp = "foo"
dummy_args.experiment_path = args.experiment_path
if args.encoder_loss.lower() == "bce":
tag = "BCE"
else:
tag = "MSE"
if emb == "vae":
model = Global.dataset_reference_vaes[args.dataset][0]()
home_path = Models.get_ref_model_path(dummy_args,
model.__class__.__name__,
D164.name,