def resnet34(pretrained=False, **kwargs):
"""Constructs a ResNet-34 model.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
"""
model = DepthRefineNet(BasicBlock, [3, 4, 6, 3], UpProj_Block, **kwargs)
if pretrained:
pretrained_dict = model_zoo.load_url(model_urls['resnet34'])
model.load_state_dict(update_model.update_model(model, pretrained_dict))
return model
def resnet152(pretrained=False, **kwargs):
"""Constructs a ResNet-152 model.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
"""
model = DepthRefineNet(Bottleneck, [3, 8, 36, 3], UpProj_Block, **kwargs)
if pretrained:
print('==> Load pretrained model..')
pretrained_dict = model_zoo.load_url(model_urls['resnet152'])
model.load_state_dict(update_model.update_model(model, pretrained_dict))
return model
def resnet50(pretrained=False, **kwargs):
"""Constructs a ResNet-50 model.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
"""
model = ResNet(Bottleneck, [3, 4, 6, 3], UpProj_Block, **kwargs)
if pretrained:
print('==> Load pretrained model..')
pretrained_dict = torch.load(model_path['resnet50'])
model.load_state_dict(update_model.update_model(model, pretrained_dict))
return model
net = model.resnet50(pretrained=args.pretrain, cspn_config=cspn_config)
elif args.data_set == 'kitti':
net = model.resnet18(pretrained=args.pretrain, cspn_config=cspn_config)
else:
print("==> input unknow dataset..")
if args.resume:
# Load best model checkpoint.
print('==> Resuming from best model..')
best_model_path = os.path.join(args.best_model_dir, 'best_model.pth')
print(best_model_path)
assert os.path.isdir(
args.best_model_dir), 'Error: no checkpoint directory found!'
best_model_dict = torch.load(best_model_path)
best_model_dict = update_model.remove_moudle(best_model_dict)
net.load_state_dict(update_model.update_model(net, best_model_dict))
if use_cuda:
net.cuda()
assert torch.cuda.device_count() == 1, 'only support single gpu'
net = torch.nn.DataParallel(net,
device_ids=range(torch.cuda.device_count()))
cudnn.benchmark = True
criterion = my_loss.Wighted_L1_Loss().cuda()
optimizer = optim.SGD(net.parameters(),
lr=args.lr,
momentum=args.momentum,
weight_decay=args.weight_decay,
nesterov=args.nesterov,
dampening=args.dampening)
data_xk = approx_kernel(kernel_structure, data_X, data_Y)
## Create a Pandas structure to store the paremeters
#dataframe test
parameters = {
'CC': [CC],
'CC2': [CC2],
'CR': [CR],
'CR2': [CR2],
'eps': [eps],
'maxeva': [maxeva],
'u': [u],
'repetitions': [repetitions],
'phi': [phi],
'sliv': [sliv]
}
parameters = pd.DataFrame(parameters)
model = create_model(parameters, data1_X, data1_Y)
#pdb.set_trace()
acc, outclass, fp, fn, answers = classify(model, data2_X, data2_Y, parameters)
batch_size = 10
model_updated = update_model(parameters, data2_X, data2_Y, batch_size, model)
pdb.set_trace()
#dataframe=dataframe.append(kernel_structure,ignore_index=True)
#access elements from df
#print(dataframe.iloc[0].loc['CC'])
#print(dataframe.iloc[1].loc['kernel_type'])