Project dependencies may have API risk issues

[PyDeps]

Hi, In MedicalNet, inappropriate dependency versioning constraints can cause risks.

Below are the dependencies and version constraints that the project is using

pip>=9.0.1
torch==0.4.1
numpy==1.15.4
nibabel==2.4.1
scipy==1.1.0
argparse==1.1

The version constraint == will introduce the risk of dependency conflicts because the scope of dependencies is too strict.
The version constraint No Upper Bound and * will introduce the risk of the missing API Error because the latest version of the dependencies may remove some APIs.

After further analysis, in this project,
The version constraint of dependency scipy can be changed to >=0.9.0,<=1.7.3.
The version constraint of dependency argparse can be changed to >=1.2.1,<=1.4.0.

The above modification suggestions can reduce the dependency conflicts as much as possible,
and introduce the latest version as much as possible without calling Error in the projects.

The invocation of the current project includes all the following methods.

The calling methods from the scipy

scipy.ndimage.interpolation.zoom

The calling methods from the argparse

argparse.ArgumentParser.parse_args
argparse.ArgumentParser

The calling methods from the all methods

torch.load
self.bn1
self.__testing_data_process__
models.resnet.resnet101
self.__resize_data__.get_data
torch.nn.DataParallel.load_state_dict
os.path.isfile
self.conv3
models.resnet.resnet10
self.bn3
m.bias.data.zero_
self.layer1
self.conv2
torch.optim.SGD
torch.optim.lr_scheduler.ExponentialLR
enumerate
torch.cat
torch.nn.functional.avg_pool3d
self.modules
str
list
argparse.ArgumentParser
self.BasicBlock.super.__init__
torch.nn.MaxPool3d
isinstance
self.conv_seg
torch.nn.Sequential
self.__nii2tensorarray__
torch.nn.DataParallel.cuda
new_parameters.append
self.relu
setting.parse_opts
filter
train
layers.append
models.resnet.resnet34
nibabel.load
new_data.astype.astype
model.load_state_dict
volumes.cuda.cuda
self.layer2
torch.nn.ReLU
torch.nn.BatchNorm3d
loss_seg.cuda.cuda
loss.backward
torch.nn.DataParallel.parameters
self.layer3
self.ResNet.super.__init__
line.strip
self.__random_center_crop__
torch.optim.SGD.zero_grad
torch.nn.DataParallel.named_parameters
format
len
self._make_layer
models.resnet.resnet50
self.conv1
out.size.out.size.out.size.out.size.planes.out.size.torch.Tensor.zero_
datasets.brains18.BrainS18Dataset
functools.partial
os.makedirs
models.resnet.resnet152
random.random
numpy.random.normal
scipy.ndimage.interpolation.zoom
numpy.reshape
torch.nn.DataParallel
logging.getLogger
exit
ResNet
model.state_dict
torch.utils.data.DataLoader
self.__training_data_process__
numpy.where
m.weight.data.fill_
self.Bottleneck.super.__init__
new_label_masks.cuda.cuda
model.state_dict.keys
print
block
os.path.join
logging.basicConfig
models.resnet.resnet200
torch.optim.SGD.step
os.path.exists
range
models.resnet.resnet18
model.state_dict.update
argparse.ArgumentParser.add_argument
self.__crop_data__
pname.find
numpy.array
numpy.min
self.__resize_data__
torch.nn.Conv3d
self.layer4
open
pixels.std
torch.Tensor
os.path.dirname
self.relu.size
self.bn2
model.generate_model
torch.nn.CrossEntropyLoss
model
map
torch.autograd.Variable
argparse.ArgumentParser.set_defaults
torch.nn.init.kaiming_normal
loss.item
torch.save
torch.optim.lr_scheduler.ExponentialLR.get_lr
numpy.max
numpy.zeros
torch.nn.DataParallel.state_dict
super
torch.optim.SGD.state_dict
new_label_masks.torch.tensor.to
torch.optim.SGD.load_state_dict
torch.manual_seed
self.__itensity_normalize_one_volume__.get_data
model.train
loss_seg
torch.load.items
time.time
argparse.ArgumentParser.parse_args
id
fio.read
self.maxpool
torch.nn.ConvTranspose3d
self.downsample
torch.tensor
torch.optim.lr_scheduler.ExponentialLR.step
fio.read.splitlines
loss_seg.item
utils.logger.log.info
idx.self.img_list.split
int
self.__itensity_normalize_one_volume__
conv3x3x3
pixels.mean
zero_pads.cuda.cuda
self.__drop_invalid_range__

@developer
Could please help me check this issue?
May I pull a request to fix it?
Thank you very much.