num_workers=n_cpu,
drop_last=True)
# dataset.__len__() : 47 (dataset/bs)
# the 8 golf swing events are classes 0 through 7, no-event is class 8
# the ratio of events to no-events is approximately 1:35 so weight classes accordingly:
# TODO: edit weights shape from golf-8-element to stsq-12-element
if use_no_element == False:
weights = torch.FloatTensor([1/3, 1, 2/5, 1/3, 1/6, 1, 1/4, 1, 1/4, 1/3, 1/2, 1/6]).to(device)
else:
weights = torch.FloatTensor([1/3, 1, 2/5, 1/3, 1/6, 1, 1/4, 1, 1/4, 1/3, 1/2, 1/6, 1/60]).to(device)
criterion = torch.nn.CrossEntropyLoss(weight=weights)
optimizer = torch.optim.Adam(filter(lambda p: p.requires_grad, model.parameters()), lr=0.001) ##lambda:無名関数
losses = AverageMeter()
#print('utils.py, class AverageMeter()')
if not os.path.exists('models'):
os.mkdir('models')
epoch = 0
for epoch in range(int(iterations)):
# while i < int(iterations):
for sample in tqdm(data_loader):
images, labels = sample['images'].to(device), sample['labels'].to(device)
logits = model(images)
seq_length=seq_length,
train=True)
data_loader = DataLoader(dataset,
batch_size=bs,
shuffle=True,
num_workers=n_cpu,
drop_last=True)
# the 8 golf swing events are classes 0 through 7, no-event is class 8
# the ratio of events to no-events is approximately 1:35 so weight classes accordingly:
weights = torch.FloatTensor(
[1/8, 1/8, 1/8, 1/8, 1/8, 1/8, 1/8, 1/8, 1/35]).cuda()
criterion = torch.nn.CrossEntropyLoss(weight=weights)
optimizer = torch.optim.Adam(
filter(lambda p: p.requires_grad, model.parameters()), lr=0.001)
losses = AverageMeter()
# writer = SummaryWriter()
if not os.path.exists('models'):
os.mkdir('models')
i = 0
while i < iterations:
# for p in optimizer.param_groups:
# print(p['lr'])
for sample in data_loader:
images, labels = sample['images'].cuda(), sample['labels'].cuda()
logits = model(images)
noise_level=noise_level)
data_loader = DataLoader(dataset,
batch_size=bs,
shuffle=True,
num_workers=n_cpu,
drop_last=True)
# the 8 golf swing events are classes 0 through 7, no-event is class 8
# the ratio of events to no-events is approximately 1:35 so weight classes accordingly:
weights = torch.FloatTensor(
[1 / 8, 1 / 8, 1 / 8, 1 / 8, 1 / 8, 1 / 8, 1 / 8, 1 / 8,
1 / 35]).cuda()
criterion = torch.nn.CrossEntropyLoss(weight=weights)
optimizer = torch.optim.Adam(filter(lambda p: p.requires_grad,
model.parameters()),
lr=0.001)
losses = AverageMeter()
if not os.path.exists('models'):
os.mkdir('models')
i = 0
while i < iterations:
for sample in data_loader:
images, labels = sample['images'].cuda(), sample['labels'].cuda()
logits = model(images)
Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225])]),
train=True,
noise_level=noise_level)
data_loader = DataLoader(dataset,
batch_size=bs,
shuffle=True,
num_workers=n_cpu,
drop_last=True)
# the 8 golf swing events are classes 0 through 7, no-event is class 8
# the ratio of events to no-events is approximately 1:35 so weight classes accordingly:
weights = torch.FloatTensor([1/8, 1/8, 1/8, 1/8, 1/8, 1/8, 1/8, 1/8, 1/35]).to(device)
criterion = torch.nn.CrossEntropyLoss(weight=weights)
if config['Adam']:
optimizer = torch.optim.Adam(filter(lambda p: p.requires_grad, model.parameters()), lr=config['learning_rate']) #todo(default):0.001
else:
optimizer = torch.optim.SGD(filter(lambda p: p.requires_grad, model.parameters()), lr=config['learning_rate']) #todo(default):0.001
losses = AverageMeter()
if not os.path.exists('models'):
os.mkdir('models')
i = 0
while i < iterations:
for sample in data_loader:
images, labels = sample['images'].to(device), sample['labels'].to(device)
logits = model(images)
if bool_classical_loss: