From c13dba86b2dbe581353b72602d7fa6e40991964c Mon Sep 17 00:00:00 2001
From: natanielruiz <nataniel777@hotmail.com>
Date: 星期三, 27 九月 2017 04:11:23 +0800
Subject: [PATCH] next
---
code/train.py | 84 ++++++++++++++++++++++++++++--------------
1 files changed, 56 insertions(+), 28 deletions(-)
diff --git a/code/train.py b/code/train.py
index ff060af..3525f87 100644
--- a/code/train.py
+++ b/code/train.py
@@ -46,6 +46,9 @@
parser.add_argument('--output_string', dest='output_string', help='String appended to output snapshots.', default = '', type=str)
parser.add_argument('--alpha', dest='alpha', help='Regression loss coefficient.',
default=0.001, type=float)
+ parser.add_argument('--iter_ref', dest='iter_ref', help='Number of iterative refinement passes.',
+ default=1, type=int)
+ parser.add_argument('--dataset', dest='dataset', help='Dataset type.', default='Pose_300W_LP', type=str)
args = parser.parse_args()
return args
@@ -111,7 +114,7 @@
# ResNet101 with 3 outputs
# model = hopenet.Hopenet(torchvision.models.resnet.Bottleneck, [3, 4, 23, 3], 66)
# ResNet50
- model = hopenet.Hopenet(torchvision.models.resnet.Bottleneck, [3, 4, 6, 3], 66)
+ model = hopenet.Hopenet(torchvision.models.resnet.Bottleneck, [3, 4, 6, 3], 66, args.iter_ref)
# ResNet18
# model = hopenet.Hopenet(torchvision.models.resnet.BasicBlock, [2, 2, 2, 2], 66)
load_filtered_state_dict(model, model_zoo.load_url(model_urls['resnet50']))
@@ -122,8 +125,21 @@
transforms.RandomCrop(224), transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])])
- pose_dataset = datasets.Pose_300W_LP(args.data_dir, args.filename_list,
- transformations)
+ if args.dataset == 'Pose_300W_LP':
+ pose_dataset = datasets.Pose_300W_LP(args.data_dir, args.filename_list, transformations)
+ elif args.dataset == 'AFLW2000':
+ pose_dataset = datasets.AFLW2000(args.data_dir, args.filename_list, transformations)
+ elif args.dataset == 'BIWI':
+ pose_dataset = datasets.BIWI(args.data_dir, args.filename_list, transformations)
+ elif args.dataset == 'AFLW':
+ pose_dataset = datasets.AFLW(args.data_dir, args.filename_list, transformations)
+ elif args.dataset == 'AFLW_aug':
+ pose_dataset = datasets.AFLW_aug(args.data_dir, args.filename_list, transformations)
+ elif args.dataset == 'AFW':
+ pose_dataset = datasets.AFW(args.data_dir, args.filename_list, transformations)
+ else:
+ print 'Error: not a valid dataset name'
+ sys.exit()
train_loader = torch.utils.data.DataLoader(dataset=pose_dataset,
batch_size=batch_size,
shuffle=True,
@@ -133,6 +149,7 @@
softmax = nn.Softmax()
criterion = nn.CrossEntropyLoss().cuda()
reg_criterion = nn.MSELoss().cuda()
+ smooth_l1_loss = nn.SmoothL1Loss().cuda()
# Regression loss coefficient
alpha = args.alpha
@@ -141,18 +158,23 @@
optimizer = torch.optim.Adam([{'params': get_ignored_params(model), 'lr': 0},
{'params': get_non_ignored_params(model), 'lr': args.lr},
- {'params': get_fc_params(model), 'lr': args.lr * 2}],
+ {'params': get_fc_params(model), 'lr': args.lr * 5}],
lr = args.lr)
print 'Ready to train network.'
print 'First phase of training.'
for epoch in range(num_epochs):
- for i, (images, labels, name) in enumerate(train_loader):
+ for i, (images, labels, cont_labels, name) in enumerate(train_loader):
images = Variable(images.cuda(gpu))
label_yaw = Variable(labels[:,0].cuda(gpu))
label_pitch = Variable(labels[:,1].cuda(gpu))
label_roll = Variable(labels[:,2].cuda(gpu))
+
+ label_angles = Variable(cont_labels[:,:3].cuda(gpu))
+ label_yaw_cont = Variable(cont_labels[:,0].cuda(gpu))
+ label_pitch_cont = Variable(cont_labels[:,1].cuda(gpu))
+ label_roll_cont = Variable(cont_labels[:,2].cuda(gpu))
optimizer.zero_grad()
model.zero_grad()
@@ -169,22 +191,20 @@
pitch_predicted = softmax(pre_pitch)
roll_predicted = softmax(pre_roll)
- yaw_predicted = torch.sum(yaw_predicted * idx_tensor, 1)
- pitch_predicted = torch.sum(pitch_predicted * idx_tensor, 1)
- roll_predicted = torch.sum(roll_predicted * idx_tensor, 1)
+ yaw_predicted = torch.sum(yaw_predicted * idx_tensor, 1) * 3 - 99
+ pitch_predicted = torch.sum(pitch_predicted * idx_tensor, 1) * 3 - 99
+ roll_predicted = torch.sum(roll_predicted * idx_tensor, 1) * 3 - 99
- loss_reg_yaw = reg_criterion(yaw_predicted, label_yaw.float())
- loss_reg_pitch = reg_criterion(pitch_predicted, label_pitch.float())
- loss_reg_roll = reg_criterion(roll_predicted, label_roll.float())
+ loss_reg_yaw = reg_criterion(yaw_predicted, label_yaw_cont)
+ loss_reg_pitch = reg_criterion(pitch_predicted, label_pitch_cont)
+ loss_reg_roll = reg_criterion(roll_predicted, label_roll_cont)
- # print yaw_predicted, label_yaw.float(), loss_reg_yaw
# Total loss
loss_yaw += alpha * loss_reg_yaw
loss_pitch += alpha * loss_reg_pitch
loss_roll += alpha * loss_reg_roll
loss_seq = [loss_yaw, loss_pitch, loss_roll]
- # loss_seq = [loss_reg_yaw, loss_reg_pitch, loss_reg_roll]
grad_seq = [torch.Tensor(1).cuda(gpu) for _ in range(len(loss_seq))]
torch.autograd.backward(loss_seq, grad_seq)
optimizer.step()
@@ -204,12 +224,16 @@
print 'Second phase of training (finetuning layer).'
for epoch in range(num_epochs_ft):
- for i, (images, labels, name) in enumerate(train_loader):
+ for i, (images, labels, cont_labels, name) in enumerate(train_loader):
images = Variable(images.cuda(gpu))
label_yaw = Variable(labels[:,0].cuda(gpu))
label_pitch = Variable(labels[:,1].cuda(gpu))
label_roll = Variable(labels[:,2].cuda(gpu))
- label_angles = Variable(labels[:,:3].cuda(gpu))
+
+ label_angles = Variable(cont_labels[:,:3].cuda(gpu))
+ label_yaw_cont = Variable(cont_labels[:,0].cuda(gpu))
+ label_pitch_cont = Variable(cont_labels[:,1].cuda(gpu))
+ label_roll_cont = Variable(cont_labels[:,2].cuda(gpu))
optimizer.zero_grad()
model.zero_grad()
@@ -226,13 +250,13 @@
pitch_predicted = softmax(pre_pitch)
roll_predicted = softmax(pre_roll)
- yaw_predicted = torch.sum(yaw_predicted.data * idx_tensor, 1)
- pitch_predicted = torch.sum(pitch_predicted.data * idx_tensor, 1)
- roll_predicted = torch.sum(roll_predicted.data * idx_tensor, 1)
+ yaw_predicted = torch.sum(yaw_predicted * idx_tensor, 1) * 3 - 99
+ pitch_predicted = torch.sum(pitch_predicted * idx_tensor, 1) * 3 - 99
+ roll_predicted = torch.sum(roll_predicted * idx_tensor, 1) * 3 - 99
- loss_reg_yaw = reg_criterion(yaw_predicted, label_yaw.float())
- loss_reg_pitch = reg_criterion(pitch_predicted, label_pitch.float())
- loss_reg_roll = reg_criterion(roll_predicted, label_roll.float())
+ loss_reg_yaw = reg_criterion(yaw_predicted, label_yaw_cont)
+ loss_reg_pitch = reg_criterion(pitch_predicted, label_pitch_cont)
+ loss_reg_roll = reg_criterion(roll_predicted, label_roll_cont)
# Total loss
loss_yaw += alpha * loss_reg_yaw
@@ -240,9 +264,17 @@
loss_roll += alpha * loss_reg_roll
# Finetuning loss
- loss_angles = reg_criterion(angles[0], label_angles.float())
+ loss_seq = [loss_yaw, loss_pitch, loss_roll]
+ for idx in xrange(1,len(angles)):
+ label_angles_residuals = label_angles - angles[0] * 3 - 99
+ for idy in xrange(1,idx):
+ label_angles_residuals += angles[idy] * 3 - 99
+ label_angles_residuals = label_angles_residuals.detach()
+ # Reconvert to other unit
+ label_angles_residuals = label_angles_residuals / 3.0 + 33
+ loss_angles = smooth_l1_loss(angles[idx], label_angles_residuals)
+ loss_seq.append(loss_angles)
- loss_seq = [loss_yaw, loss_pitch, loss_roll, loss_angles]
grad_seq = [torch.Tensor(1).cuda(gpu) for _ in range(len(loss_seq))]
torch.autograd.backward(loss_seq, grad_seq)
optimizer.step()
@@ -255,11 +287,7 @@
# 'output/snapshots/' + args.output_string + '_iter_'+ str(i+1) + '.pkl')
# Save models at numbered epochs.
- if epoch % 1 == 0 and epoch < num_epochs_ft - 1:
+ if epoch % 1 == 0 and epoch < num_epochs_ft:
print 'Taking snapshot...'
torch.save(model.state_dict(),
'output/snapshots/' + args.output_string + '_epoch_'+ str(num_epochs+epoch+1) + '.pkl')
-
-
- # Save the final Trained Model
- torch.save(model.state_dict(), 'output/snapshots/' + args.output_string + '_epoch_' + str(num_epochs+epoch+1) + '.pkl')
--
Gitblit v1.8.0