def train(epoch, model, criterion, optimizer, train_loader, args):
losses = AverageMeter()
batch_time = AverageMeter()
accuracy = AverageMeter()
pos_sims = AverageMeter()
neg_sims = AverageMeter()
end = time.time()
freq = min(args.print_freq, len(train_loader))
for i, data_ in enumerate(train_loader, 0):
inputs, labels = data_
#print(inputs)
# wrap them in Variable
inputs = Variable(inputs).cuda()
labels = Variable(labels).cuda()
optimizer.zero_grad()
embed_feat = model(inputs)
loss = criterion(embed_feat, labels)
#get_dot = bad_grad_viz.register_hooks(loss)
if args.orth_reg != 0:
loss = orth_reg(net=model, loss=loss, cof=args.orth_reg)
loss.backward()
optimizer.step()
#for param in model.parameters():
# print(param.grad.size())
#dot = get_dot()
#dot.save('tmp.dot')
# measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
losses.update(loss.item())
#accuracy.update(inter_)
#pos_sims.update(dist_ap)
#neg_sims.update(dist_an)
if (i + 1) % freq == 0 or (i + 1) == len(train_loader):
print('Epoch: [{0:03d}][{1}/{2}]\t'
'Time {batch_time.avg:.3f}\t'
'Loss {loss.avg:.8f} \t'.format(epoch + 1,
i + 1,
len(train_loader),
batch_time=batch_time,
loss=losses))
if epoch == 0 and i == 0:
print('-- HA-HA-HA-HA-AH-AH-AH-AH --')
def train(epoch, model, criterion, optimizer, train_loader, args):
losses = AverageMeter()
batch_time = AverageMeter()
accuracy = AverageMeter()
pos_sims = AverageMeter()
neg_sims = AverageMeter()
end = time.time()
freq = min(args.print_freq, len(train_loader))
for i, data_ in enumerate(train_loader, 0):
inputs, labels = data_
# wrap them in Variable
inputs = Variable(inputs).cuda()
labels = Variable(labels).cuda()
optimizer.zero_grad()
embed_feat = model(inputs)
loss, inter_, dist_ap, dist_an = criterion(embed_feat, labels)
if args.orth_reg != 0:
loss = orth_reg(net=model, loss=loss, cof=args.orth_reg)
loss.backward()
optimizer.step()
# measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
losses.update(loss.item())
accuracy.update(inter_)
pos_sims.update(dist_ap)
neg_sims.update(dist_an)
if (i + 1) % freq == 0 or (i + 1) == len(train_loader):
print('Epoch: [{0:03d}][{1}/{2}]\t'
'Time {batch_time.avg:.3f}\t'
'Loss {loss.avg:.4f} \t'
'Accuracy {accuracy.avg:.4f} \t'
'Pos {pos.avg:.4f}\t'
'Neg {neg.avg:.4f} \t'.format(epoch + 1,
i + 1,
len(train_loader),
batch_time=batch_time,
loss=losses,
accuracy=accuracy,
pos=pos_sims,
neg=neg_sims))
if epoch == 0 and i == 0:
print('-- HA-HA-HA-HA-AH-AH-AH-AH --')
def train(epoch, model, criterion, optimizer, train_loader, args):
losses = AverageMeter()
batch_time = AverageMeter()
accuracy = AverageMeter()
pos_sims = AverageMeter()
neg_sims = AverageMeter()
end = time.time()
freq = min(args.print_freq, len(train_loader))
for i, data_ in enumerate(train_loader, 0):
inputs, poss, negs, labels = data_
inputs = inputs.cuda()
labels = labels.cuda()
poss = poss.cuda()
negs = negs.cuda()
optimizer.zero_grad()
anchor_emb = model(inputs)
poss_emb = model(poss)
negs_emb = model(negs)
loss = criterion(anchor_emb, poss_emb, negs_emb)
if args.orth_reg != 0:
loss = orth_reg(net=model, loss=loss, cof=args.orth_reg)
loss.backward()
optimizer.step()
# measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
losses.update(loss.data.item())
if (i + 1) % freq == 0 or (i+1) == len(train_loader):
print('Epoch: [{0:03d}][{1}/{2}]\t'
'Time {batch_time.avg:.3f}\t'
'Loss {loss.avg:.4f} \t'.format
(epoch + 1, i + 1, len(train_loader), batch_time=batch_time,
loss=losses,))
if epoch == 0 and i == 0:
print('-- HA-HA-HA-HA-AH-AH-AH-AH --')
# get the inputs
inputs, labels = data
# break
# wrap them in Variable
inputs = Variable(inputs.cuda())
labels = Variable(labels).cuda()
# zero the parameter gradients
optimizer.zero_grad()
# forward + backward + optimize
embed_feat = model(inputs)
# loss = criterion(embed_feat, labels)
loss, inter_, dist_ap, dist_an = criterion(embed_feat, labels)
if args.orth_cof > 1e-9:
loss = orth_reg(model, loss, cof=args.orth_cof)
loss.backward()
optimizer.step()
running_loss += loss.data[0]
# print(epoch)
print(
'[epoch %05d]\t loss: %.7f \t prec: %.3f \t pos-dist: %.3f \tneg-dist: %.3f'
% (epoch + 1, running_loss, inter_, dist_ap, dist_an))
if epoch % args.save_step == 0:
torch.save(model, os.path.join(log_dir, '%d_model.pkl' % epoch))
torch.save(model, os.path.join(log_dir, '%d_model.pkl' % epoch))
print('Finished Training')
def train(epoch, model, criterion, optimizer, train_loader, args):
losses = AverageMeter()
batch_time = AverageMeter()
accuracy = AverageMeter()
pos_sims = AverageMeter()
neg_sims = AverageMeter()
end = time.time()
freq = min(args.print_freq, len(train_loader))
if not args.use_test or True:
test_loader = [(0, 0) for _ in range(len(train_loader))]
else:
test_loader = args.use_test
for i, (data_, test_data_) in enumerate(zip(train_loader, test_loader), 0):
inputs, labels = data_
inputs_test, _ = test_data_
num_samples, _, w, h = inputs.size()
inputs_1 = inputs[:, 0:3, :, :]
inputs_2 = inputs[np.random.choice(range(num_samples), args.rot_batch), :, :, :].view(-1, 3, w, h)
if args.use_test and False:
num_samples = inputs_test.size(0)
inputs_3 = inputs_test[np.random.choice(range(num_samples), args.rot_batch), :, :, :].view(-1, 3, w, h)
inputs_3 = Variable(inputs_3).cuda()
# wrap them in Variable
inputs_1 = Variable(inputs_1).cuda()
inputs_2 = Variable(inputs_2).cuda()
labels = Variable(labels).cuda()
optimizer.zero_grad()
if not args.rot_only:
embed_feat = model(inputs_1, rot=False)
if args.dim % 64 != 0:
loss, inter_, dist_ap, dist_an = nn.CrossEntropyLoss()(embed_feat, labels), 0, 0, 0
else:
loss, inter_, dist_ap, dist_an = criterion(embed_feat, labels)
else:
loss, inter_, dist_ap, dist_an = 0, 0, 0, 0
loss_rot = torch.zeros(1)
loss_rot_test = torch.zeros(1)
if args.self_supervision_rot:
score = model(inputs_2, rot=True)
labels_rot = torch.LongTensor([0, 1, 2, 3] * args.rot_batch).cuda()
loss_rot = nn.CrossEntropyLoss()(score, labels_rot)
loss += args.self_supervision_rot * loss_rot
if args.orth_reg != 0:
loss = orth_reg(net=model, loss=loss, cof=args.orth_reg)
loss.backward()
optimizer.step()
# measure elapsed time
batch_time.update(time.time() - end)
end = time.time()
if not args.rot_only:
losses.update(loss.item())
accuracy.update(inter_)
pos_sims.update(dist_ap)
neg_sims.update(dist_an)
if (i + 1) % freq == 0 or (i+1) == len(train_loader):
print('Epoch: [{0:03d}][{1}/{2}]\t'
'Time {batch_time.avg:.3f}\t'
'Loss {loss.avg:.4f} \t'
'Loss_rot {loss_rot:.4f} \t'
'Loss_rot_test {loss_rot_test:.4f} \t'
'accuracy {accuracy.avg:.4f} \t'
'Pos {pos.avg:.4f}\t'
'Neg {neg.avg:.4f} \t'.format
(epoch + 1, i + 1, len(train_loader), batch_time=batch_time,
loss=losses, loss_rot=loss_rot.item(), loss_rot_test=loss_rot_test.item(), accuracy=accuracy, pos=pos_sims, neg=neg_sims))
def main(args):
# 训练日志保存
log_dir = os.path.join('checkpoints', args.log_dir)
mkdir_if_missing(log_dir)
sys.stdout = logging.Logger(os.path.join(log_dir, 'log.txt'))
display(args)
if args.r is None:
model = models.create(args.net, Embed_dim=args.dim)
# load part of the model
model_dict = model.state_dict()
# print(model_dict)
if args.net == 'bn':
pretrained_dict = torch.load(
'pretrained_models/bn_inception-239d2248.pth')
else:
pretrained_dict = torch.load(
'pretrained_models/inception_v3_google-1a9a5a14.pth')
pretrained_dict = {
k: v
for k, v in pretrained_dict.items() if k in model_dict
}
model_dict.update(pretrained_dict)
# orth init
if args.init == 'orth':
print('initialize the FC layer orthogonally')
_, _, v = torch.svd(model_dict['Embed.linear.weight'])
model_dict['Embed.linear.weight'] = v.t()
# zero bias
model_dict['Embed.linear.bias'] = torch.zeros(args.dim)
model.load_state_dict(model_dict)
else:
# resume model
model = torch.load(args.r)
model = model.cuda()
torch.save(model, os.path.join(log_dir, 'model.pkl'))
print('initial model is save at %s' % log_dir)
# fine tune the model: the learning rate for pre-trained parameter is 1/10
new_param_ids = set(map(id, model.Embed.parameters()))
new_params = [p for p in model.parameters() if id(p) in new_param_ids]
base_params = [p for p in model.parameters() if id(p) not in new_param_ids]
param_groups = [{
'params': base_params,
'lr_mult': 0.1
}, {
'params': new_params,
'lr_mult': 1.0
}]
optimizer = torch.optim.Adam(param_groups,
lr=args.lr,
weight_decay=args.weight_decay)
criterion = losses.create(args.loss, alpha=args.alpha, k=args.k).cuda()
data = DataSet.create(args.data, root=None, test=False)
train_loader = torch.utils.data.DataLoader(
data.train,
batch_size=args.BatchSize,
sampler=RandomIdentitySampler(data.train,
num_instances=args.num_instances),
drop_last=True,
num_workers=args.nThreads)
for epoch in range(args.start, args.epochs):
running_loss = 0.0
for i, data in enumerate(train_loader, 0):
inputs, labels = data
# wrap them in Variable
inputs = Variable(inputs.cuda())
labels = Variable(labels).cuda()
optimizer.zero_grad()
embed_feat = model(inputs)
loss, inter_, dist_ap, dist_an = criterion(embed_feat, labels)
if args.orth > 0:
loss = orth_reg(model, loss, cof=args.orth)
loss.backward()
optimizer.step()
running_loss += loss.data[0]
if epoch == 0 and i == 0:
print(50 * '#')
print('Train Begin -- HA-HA-HA')
print(
'[Epoch %05d]\t Loss: %.3f \t Accuracy: %.3f \t Pos-Dist: %.3f \t Neg-Dist: %.3f'
% (epoch + 1, running_loss, inter_, dist_ap, dist_an))
if epoch % args.save_step == 0:
torch.save(model, os.path.join(log_dir, '%d_model.pkl' % epoch))