def saveModels(self):
torch.save(
self.generator,
util.fcnt(self.checkpoints_dir, "voxel_flow_generator", "pth"))
torch.save(
self.discriminator,
util.fcnt(self.checkpoints_dir, "voxel_flow_discriminator", "pth"))
print("done save models")
def saveModels(self):
name = util.fcnt(self.sheckpoints_dir, "pix2pix_generator", "pth")
torch.save(self.generator, name)
print("{} is saved".format(name))
name = util.fcnt(self.sheckpoints_dir, "pix2pix_discriminator", "pth")
torch.save(self.discriminator, name)
print("{} is saved".format(name))
# torch.save(self.discriminator, util.fcnt(self.sheckpoints_dir, "pix2pix_discriminator", "pth"))
print("done save models")
def save_models(self):
name = util.fcnt(self.checkpoints_dir,
"annotation_controller_generator", "pth")
torch.save(self.generator, name)
print("{} is saved".format(name))
name = util.fcnt(self.checkpoints_dir,
"annotation_controller_discriminator", "pth")
torch.save(self.discriminator, name)
print("{} is saved".format(name))
# torch.save(self.discriminator, util.fcnt(self.checkpoints_dir, "annotation controller_discriminator", "pth"))
print("done save models")
def _saveModel(self, model, file_name, is_fcnt=True):
model_name = str(model.__class__).split("'")[1]
if model_name.split(".")[-1] == "BaseSequential":
self._saveModel(list(model)[0], file_name)
else:
if is_fcnt:
path = util.fcnt(self.sheckpoints_dir, file_name, "pth")
else:
path = file_name
state_dict = model.state_dict()
state_dict['model_name'] = model_name
state_dict['setting'] = self.setting
torch.save(state_dict, path)
print("{} is saved".format(path))
def dump(self, image=None):
img = I.fromarray(image[...,::-1])
img.save(util.fcnt(dir=self.dump_path, fname = "image", ext=self.ext))