def get_model():
# Get model from config
if config.model == "resnet18":
model = models.resnet18(pretrained=config.pretrained)
elif config.model == "resnet34":
model = models.resnet34(pretrained=config.pretrained)
elif config.model == 'resnet50':
model = models.resnet50(pretrained=config.pretrained)
elif config.model == "resnet101":
model = models.resnet101(pretrained=config.pretrained)
elif config.model == "resnet152":
model = models.resnet152(pretrained=config.pretrained)
elif config.model == "resnext50_32x4d":
model = models.resnet34(pretrained=config.pretrained)
elif config.model == 'resnext101_32x8d':
model = models.resnet50(pretrained=config.pretrained)
elif config.model == "wide_resnet50_2":
model = models.resnet101(pretrained=config.pretrained)
elif config.model == "wide_resnet101_2":
model = models.resnet152(pretrained=config.pretrained)
else:
raise ValueError('%s not supported'.format(config.model))
# Initialize fc layer
(in_features, out_features) = model.fc.in_features, model.fc.out_features
model.fc = torch.nn.Linear(in_features, out_features)
return model
def __init__(self, num_categories):
super(CategoryModel, self).__init__()
resnet101 = models.resnet101(pretrained=True)
# resnet18 = models.resnet18(pretrained=True)
self.model = nn.Sequential(*list(resnet101.children())[:-2])
self.avgpool = nn.AdaptiveAvgPool2d(output_size=(1, 1))
self.inp_features = resnet101.fc.in_features
self.fc = nn.Linear(self.inp_features, num_categories)
def __init__(self, checkpoint):
self.checkpoint = torch.load(checkpoint, map_location=device)
print('loading model: {}...'.format(checkpoint))
# self.model = self.checkpoint['model'].module
self.args = parse_args()
self.model = resnet101(self.args)
self.model.load_state_dict(self.checkpoint)
self.model.to(device)
self.model.eval()
self.transformer = data_transforms['val']
def Resnet101(num_classes, test=False):
model = resnet101()
if not test:
if LOCAL_PRETRAINED['resnet101'] == None:
state_dict = load_state_dict_from_url(model_urls['resnet101'],
progress=True)
else:
state_dict = state_dict = torch.load(LOCAL_PRETRAINED['resnet101'])
model.load_state_dict(state_dict)
fc_features = model.fc.in_features
model.fc = nn.Linear(fc_features, num_classes)
return model
def initModel(args):
# Setup Model
if args.arch == "resnet50":
model = models.resnet50(pretrained=True,
num_classes=18,
scale=args.scale)
elif args.arch == "resnet101":
model = models.resnet101(pretrained=True,
num_classes=18,
scale=args.scale)
elif args.arch == "resnet152":
model = models.resnet152(pretrained=True,
num_classes=18,
scale=args.scale)
elif args.arch == "vgg16":
model = models.vgg16(pretrained=True, num_classes=18)
elif args.arch == "googlenet":
model = models.googlenet(pretrained=True, num_classes=18)
for param in model.parameters():
param.requires_grad = False
#model = model.cuda()
if args.resume is not None:
if os.path.isfile(args.resume):
print(("Loading model and optimizer from checkpoint '{}'".format(
args.resume)))
checkpoint = torch.load(args.resume,
map_location=torch.device('cpu'))
# model.load_state_dict(checkpoint['state_dict'])
d = collections.OrderedDict()
for key, value in list(checkpoint['state_dict'].items()):
tmp = key[7:]
d[tmp] = value
model.load_state_dict(d)
print(("Loaded checkpoint '{}' (epoch {})".format(
args.resume, checkpoint['epoch'])))
sys.stdout.flush()
else:
print(("No checkpoint found at '{}'".format(args.resume)))
sys.stdout.flush()
model.eval()
summary(model, (3, 640, 640))
return model
def test(args):
data_loader = IC15TestLoader(long_size=args.long_size)
test_loader = torch.utils.data.DataLoader(
data_loader,
batch_size=1,
shuffle=False,
num_workers=2,
drop_last=True)
submit_path = 'outputs'
if os.path.exists(submit_path):
shutil.rmtree(submit_path)
os.mkdir(submit_path)
# Setup Model
if args.arch == "resnet50":
model = models.resnet50(pretrained=True, num_classes=7, scale=args.scale)
elif args.arch == "resnet101":
model = models.resnet101(pretrained=True, num_classes=7, scale=args.scale)
elif args.arch == "resnet152":
model = models.resnet152(pretrained=True, num_classes=7, scale=args.scale)
for param in model.parameters():
param.requires_grad = False
model = model.cuda()
if args.resume is not None:
if os.path.isfile(args.resume):
print("Loading model and optimizer from checkpoint '{}'".format(args.resume))
checkpoint = torch.load(args.resume)
# model.load_state_dict(checkpoint['state_dict'])
d = collections.OrderedDict()
for key, value in checkpoint['state_dict'].items():
tmp = key[7:]
d[tmp] = value
model.load_state_dict(d)
print("Loaded checkpoint '{}' (epoch {})"
.format(args.resume, checkpoint['epoch']))
sys.stdout.flush()
else:
print("No checkpoint found at '{}'".format(args.resume))
sys.stdout.flush()
state_dict = model.state_dict() #if data_parallel else model.state_dict()
torch.save(state_dict,'./ctpn_pse_batch_new_pre_train_ic15.pth')
def build_model(model_name, num_classes, pretrain):
if model_name == 'resnet50':
net = resnet50(num_classes=num_classes, pretrain=pretrain)
elif model_name == 'resnet18':
net = resnet18(num_classes=num_classes, pretrain=pretrain)
elif model_name == 'resnet34':
net = resnet34(num_classes=num_classes, pretrain=pretrain)
elif model_name == 'resnet101':
net = resnet101(num_classes=num_classes, pretrain=pretrain)
elif model_name == 'resnet152':
net = resnet152(num_classes=num_classes, pretrain=pretrain)
elif model_name == 'resnet50se':
net = resnet50se(num_classes=num_classes, pretrain=pretrain)
elif model_name == 'resnet50dilated':
net = resnet50_dilated(num_classes=num_classes, pretrain=pretrain)
elif model_name == 'resnet50dcse':
net = resnet50_dcse(num_classes=num_classes, pretrain=pretrain)
else:
print('wait a minute')
return net
def load_model(args):
epoch = 0
# Setup Model
if args.arch == "resnet50":
model = models.resnet50(pretrained=True, num_classes=7, scale=args.scale)
elif args.arch == "resnet101":
model = models.resnet101(pretrained=True, num_classes=7, scale=args.scale)
elif args.arch == "resnet152":
model = models.resnet152(pretrained=True, num_classes=7, scale=args.scale)
for param in model.parameters():
param.requires_grad = False
model = model.cuda()
if args.resume is not None:
if os.path.isfile(args.resume):
print(("Loading model and optimizer from checkpoint '{}'".format(args.resume)))
checkpoint = torch.load(args.resume)
# model.load_state_dict(checkpoint['state_dict'])
d = collections.OrderedDict()
for key, value in list(checkpoint['state_dict'].items()):
tmp = key[7:]
d[tmp] = value
model.load_state_dict(d)
epoch = checkpoint['epoch']
print(("Loaded checkpoint '{}' (epoch {})"
.format(args.resume, checkpoint['epoch'])))
sys.stdout.flush()
else:
print(("No checkpoint found at '{}'".format(args.resume)))
sys.stdout.flush()
model.eval()
return model, epoch
def main(args):
if args.checkpoint == '':
args.checkpoint = "checkpoints/ctw1500_%s_bs_%d_ep_%d" % (
args.arch, args.batch_size, args.n_epoch)
if args.pretrain:
if 'synth' in args.pretrain:
args.checkpoint += "_pretrain_synth"
else:
args.checkpoint += "_pretrain_ic17"
print('checkpoint path: %s' % args.checkpoint)
print('init lr: %.8f' % args.lr)
print('schedule: ', args.schedule)
sys.stdout.flush()
if not os.path.isdir(args.checkpoint):
os.makedirs(args.checkpoint)
kernel_num = 7
min_scale = 0.4
start_epoch = 0
data_loader = CTW1500Loader(is_transform=True,
img_size=args.img_size,
kernel_num=kernel_num,
min_scale=min_scale)
#train_loader = ctw_train_loader(data_loader, batch_size=args.batch_size)
if args.arch == "resnet50":
model = models.resnet50(pretrained=True, num_classes=kernel_num)
elif args.arch == "resnet101":
model = models.resnet101(pretrained=True, num_classes=kernel_num)
elif args.arch == "resnet152":
model = models.resnet152(pretrained=True, num_classes=kernel_num)
#resnet18 and 34 didn't inplement pretrained
elif args.arch == "resnet18":
model = models.resnet18(pretrained=False, num_classes=kernel_num)
elif args.arch == "resnet34":
model = models.resnet34(pretrained=False, num_classes=kernel_num)
elif args.arch == "mobilenetv2":
model = models.resnet152(pretrained=True, num_classes=kernel_num)
elif args.arch == "mobilenetv3large":
model = models.mobilenetv3_large(pretrained=False,
num_classes=kernel_num)
elif args.arch == "mobilenetv3small":
model = models.mobilenetv3_small(pretrained=False,
num_classes=kernel_num)
optimizer = tf.keras.optimizers.SGD(learning_rate=args.lr,
momentum=0.99,
decay=5e-4)
title = 'CTW1500'
if args.pretrain:
print('Using pretrained model.')
assert os.path.isfile(
args.pretrain), 'Error: no checkpoint directory found!'
logger = Logger(os.path.join(args.checkpoint, 'log.txt'), title=title)
logger.set_names(
['Learning Rate', 'Train Loss', 'Train Acc.', 'Train IOU.'])
elif args.resume:
print('Resuming from checkpoint.')
model.load_weights(args.resume)
logger = Logger(os.path.join(args.checkpoint, 'log.txt'),
title=title,
resume=True)
else:
print('Training from scratch.')
logger = Logger(os.path.join(args.checkpoint, 'log.txt'), title=title)
logger.set_names(
['Learning Rate', 'Train Loss', 'Train Acc.', 'Train IOU.'])
for epoch in range(start_epoch, args.n_epoch):
optimizer = get_new_optimizer(args, optimizer, epoch)
print(
'\nEpoch: [%d | %d] LR: %f' %
(epoch + 1, args.n_epoch, optimizer.get_config()['learning_rate']))
train_loader = ctw_train_loader(data_loader,
batch_size=args.batch_size)
train_loss, train_te_acc, train_ke_acc, train_te_iou, train_ke_iou = train(train_loader, model, dice_loss,\
optimizer, epoch)
model.save_weights('%s%s' % (args.checkpoint, '/model_tf/weights'))
logger.append([
optimizer.get_config()['learning_rate'], train_loss, train_te_acc,
train_te_iou
])
logger.close()
def test(args, file=None):
result = []
data_loader = DataLoader(long_size=args.long_size, file=file)
test_loader = torch.utils.data.DataLoader(data_loader,
batch_size=1,
shuffle=False,
num_workers=2,
drop_last=True)
slice = 0
# Setup Model
if args.arch == "resnet50":
model = models.resnet50(pretrained=True,
num_classes=7,
scale=args.scale)
elif args.arch == "resnet101":
model = models.resnet101(pretrained=True,
num_classes=7,
scale=args.scale)
elif args.arch == "resnet152":
model = models.resnet152(pretrained=True,
num_classes=7,
scale=args.scale)
elif args.arch == "mobilenet":
model = models.Mobilenet(pretrained=True,
num_classes=6,
scale=args.scale)
slice = -1
for param in model.parameters():
param.requires_grad = False
# model = model.cuda()
if args.resume is not None:
if os.path.isfile(args.resume):
print("Loading model and optimizer from checkpoint '{}'".format(
args.resume))
checkpoint = torch.load(args.resume)
# model.load_state_dict(checkpoint['state_dict'])
d = collections.OrderedDict()
for key, value in checkpoint['state_dict'].items():
tmp = key[7:]
d[tmp] = value
try:
model.load_state_dict(d)
except:
model.load_state_dict(checkpoint['state_dict'])
print("Loaded checkpoint '{}' (epoch {})".format(
args.resume, checkpoint['epoch']))
sys.stdout.flush()
else:
print("No checkpoint found at '{}'".format(args.resume))
sys.stdout.flush()
model.eval()
total_frame = 0.0
total_time = 0.0
for idx, (org_img, img) in enumerate(test_loader):
print('progress: %d / %d' % (idx, len(test_loader)))
sys.stdout.flush()
# img = Variable(img.cuda(), volatile=True)
org_img = org_img.numpy().astype('uint8')[0]
text_box = org_img.copy()
# torch.cuda.synchronize()
start = time.time()
# angle detection
# org_img, angle = detect_angle(org_img)
outputs = model(img)
score = torch.sigmoid(outputs[:, slice, :, :])
outputs = (torch.sign(outputs - args.binary_th) + 1) / 2
text = outputs[:, slice, :, :]
kernels = outputs
# kernels = outputs[:, 0:args.kernel_num, :, :] * text
score = score.data.cpu().numpy()[0].astype(np.float32)
text = text.data.cpu().numpy()[0].astype(np.uint8)
kernels = kernels.data.cpu().numpy()[0].astype(np.uint8)
if args.arch == 'mobilenet':
pred = pse2(kernels,
args.min_kernel_area / (args.scale * args.scale))
else:
# c++ version pse
pred = pse(kernels,
args.min_kernel_area / (args.scale * args.scale))
# python version pse
# pred = pypse(kernels, args.min_kernel_area / (args.scale * args.scale))
# scale = (org_img.shape[0] * 1.0 / pred.shape[0], org_img.shape[1] * 1.0 / pred.shape[1])
scale = (org_img.shape[1] * 1.0 / pred.shape[1],
org_img.shape[0] * 1.0 / pred.shape[0])
label = pred
label_num = np.max(label) + 1
bboxes = []
rects = []
for i in range(1, label_num):
points = np.array(np.where(label == i)).transpose((1, 0))[:, ::-1]
if points.shape[0] < args.min_area / (args.scale * args.scale):
continue
score_i = np.mean(score[label == i])
if score_i < args.min_score:
continue
rect = cv2.minAreaRect(points)
bbox = cv2.boxPoints(rect) * scale
bbox = bbox.astype('int32')
bbox = order_point(bbox)
# bbox = np.array([bbox[1], bbox[2], bbox[3], bbox[0]])
bboxes.append(bbox.reshape(-1))
rec = []
rec.append(rect[-1])
rec.append(rect[1][1] * scale[1])
rec.append(rect[1][0] * scale[0])
rec.append(rect[0][0] * scale[0])
rec.append(rect[0][1] * scale[1])
rects.append(rec)
# torch.cuda.synchronize()
end = time.time()
total_frame += 1
total_time += (end - start)
print('fps: %.2f' % (total_frame / total_time))
sys.stdout.flush()
for bbox in bboxes:
cv2.drawContours(text_box, [bbox.reshape(4, 2)], -1, (0, 255, 0),
2)
image_name = data_loader.img_paths[idx].split('/')[-1].split('.')[0]
write_result_as_txt(image_name, bboxes, 'outputs/submit_invoice/')
text_box = cv2.resize(text_box, (text.shape[1], text.shape[0]))
debug(idx, data_loader.img_paths, [[text_box]], 'data/images/tmp/')
result = crnnRec(cv2.cvtColor(org_img, cv2.COLOR_BGR2RGB), rects)
result = formatResult(result)
# cmd = 'cd %s;zip -j %s %s/*' % ('./outputs/', 'submit_invoice.zip', 'submit_invoice')
# print(cmd)
# sys.stdout.flush()
# util.cmd.Cmd(cmd)
return result
def train_net(args):
torch.manual_seed(7)
np.random.seed(7)
checkpoint = args.checkpoint
start_epoch = 0
best_acc = float('-inf')
writer = SummaryWriter()
epochs_since_improvement = 0
# Initialize / load checkpoint
if checkpoint is None:
if args.network == 'r18':
model = resnet18(args)
elif args.network == 'r34':
model = resnet34(args)
elif args.network == 'r50':
model = resnet50(args)
elif args.network == 'r101':
model = resnet101(args)
elif args.network == 'r152':
model = resnet152(args)
elif args.network == 'mobile':
from mobilenet_v2 import MobileNetV2
model = MobileNetV2()
else:
raise TypeError('network {} is not supported.'.format(
args.network))
metric_fc = ArcMarginModel(args)
if args.optimizer == 'sgd':
optimizer = torch.optim.SGD([{
'params': model.parameters()
}, {
'params': metric_fc.parameters()
}],
lr=args.lr,
momentum=args.mom,
weight_decay=args.weight_decay)
else:
optimizer = torch.optim.Adam([{
'params': model.parameters()
}, {
'params': metric_fc.parameters()
}],
lr=args.lr,
weight_decay=args.weight_decay)
else:
checkpoint = torch.load(checkpoint)
start_epoch = checkpoint['epoch'] + 1
epochs_since_improvement = checkpoint['epochs_since_improvement']
model = checkpoint['model']
metric_fc = checkpoint['metric_fc']
optimizer = checkpoint['optimizer']
model = nn.DataParallel(model)
metric_fc = nn.DataParallel(metric_fc)
# Move to GPU, if available
model = model.to(device)
metric_fc = metric_fc.to(device)
# Loss function
if args.focal_loss:
criterion = FocalLoss(gamma=args.gamma).to(device)
else:
criterion = nn.CrossEntropyLoss().to(device)
# Custom dataloaders
train_dataset = ArcFaceDataset('train')
train_loader = torch.utils.data.DataLoader(train_dataset,
batch_size=args.batch_size,
shuffle=True,
num_workers=num_workers)
# Epochs
for epoch in range(start_epoch, args.end_epoch):
# Decay learning rate if there is no improvement for 2 consecutive epochs, and terminate training after 10
if epochs_since_improvement == 10:
break
if epochs_since_improvement > 0 and epochs_since_improvement % 2 == 0:
checkpoint = 'BEST_checkpoint.tar'
checkpoint = torch.load(checkpoint)
model = checkpoint['model']
metric_fc = checkpoint['metric_fc']
optimizer = checkpoint['optimizer']
adjust_learning_rate(optimizer, 0.5)
# One epoch's training
train_loss, train_top1_accs = train(train_loader=train_loader,
model=model,
metric_fc=metric_fc,
criterion=criterion,
optimizer=optimizer,
epoch=epoch)
lr = optimizer.param_groups[0]['lr']
print('\nCurrent effective learning rate: {}\n'.format(lr))
# print('Step num: {}\n'.format(optimizer.step_num))
writer.add_scalar('model/train_loss', train_loss, epoch)
writer.add_scalar('model/train_accuracy', train_top1_accs, epoch)
writer.add_scalar('model/learning_rate', lr, epoch)
if epoch % 5 == 0:
# One epoch's validation
megaface_acc = megaface_test(model)
writer.add_scalar('model/megaface_accuracy', megaface_acc, epoch)
# Check if there was an improvement
is_best = megaface_acc > best_acc
best_acc = max(megaface_acc, best_acc)
if not is_best:
epochs_since_improvement += 1
print("\nEpochs since last improvement: %d\n" %
(epochs_since_improvement, ))
else:
epochs_since_improvement = 0
# Save checkpoint
save_checkpoint(epoch, epochs_since_improvement, model, metric_fc,
optimizer, best_acc, is_best)
def test(args):
import torch
data_loader = IC15TestLoader(root_dir=args.root_dir, long_size=args.long_size)
test_loader = torch.utils.data.DataLoader(
data_loader,
batch_size=1,
shuffle=False,
num_workers=2,
drop_last=True)
# Setup Model
if args.arch == "resnet50":
model = models.resnet50(pretrained=False, num_classes=1, scale=args.scale, train_mode=False)
elif args.arch == "resnet101":
model = models.resnet101(pretrained=True, num_classes=1, scale=args.scale)
elif args.arch == "resnet152":
model = models.resnet152(pretrained=True, num_classes=1, scale=args.scale)
for param in model.parameters():
param.requires_grad = False
if args.gpus > 0:
model = model.cuda()
if args.resume is not None:
if os.path.isfile(args.resume):
print("Loading model and optimizer from checkpoint '{}'".format(args.resume))
device = torch.device('cpu') if args.gpus < 0 else None
checkpoint = torch.load(args.resume, map_location=device)
# model.load_state_dict(checkpoint['state_dict'])
d = collections.OrderedDict()
for key, value in checkpoint['state_dict'].items():
tmp = key[7:]
d[tmp] = value
model.load_state_dict(d)
print("Loaded checkpoint '{}' (epoch {})"
.format(args.resume, checkpoint['epoch']))
sys.stdout.flush()
else:
print("No checkpoint found at '{}'".format(args.resume))
sys.stdout.flush()
model.eval()
if args.onnx:
import torch.onnx.symbolic_opset9 as onnx_symbolic
def upsample_nearest_2d(g, input, output_size):
scales = g.op('Constant', value_t=torch.Tensor([1., 1., 2., 2.]))
return g.op("Upsample", input, scales, mode_s='nearest')
onnx_symbolic = upsample_nearest_2d
dummy_input = torch.autograd.Variable(torch.randn(1, 3, 640, 640)).cpu()
torch.onnx.export(model, dummy_input, 'sanet.onnx', verbose=False,
input_names=["input"],
output_names=["gaussian_map", 'border_map'],
dynamic_axes = {'input':{0:'b', 2:'h', 3:'w'}, 'gaussian_map':{0:'b', 2:'h', 3:'w'}, 'border_map':{0:'b', 2:'h', 3:'w'}}
)
return 0
total_frame = 0.0
total_time = 0.0
for idx, (org_img, img, scale_val) in enumerate(test_loader):
print('progress: %d / %d'%(idx, len(test_loader)))
sys.stdout.flush()
if args.gpus > 0:
img = Variable(img.cuda(), volatile=True)
org_img = org_img.numpy().astype('uint8')[0]
text_box = org_img.copy()
resize_img = img.cpu().numpy().astype('uint8')[0].transpose((1, 2, 0)).copy()
if args.gpus > 0:
torch.cuda.synchronize()
start = time.time()
outputs = model(img)
infer_time = time.time()
probability_map, border_map = outputs[0].sigmoid(), outputs[1].sigmoid()
# print(probability_map.max(), probability_map.min())
score = probability_map[0, 0]
border_score = border_map[0, 0]
# prediction_map = textfill(score.cpu().numpy(), border_score, top_threshold=0.7, end_thershold=0.2)
post_time = time.time()
center_text = torch.where(score > 0.7, torch.ones_like(score), torch.zeros_like(score))
center_text = center_text.data.cpu().numpy().astype(np.uint8)
text_region = torch.where(score > 0.5, torch.ones_like(score), torch.zeros_like(score))
border_region = torch.where(border_score > 0.9, torch.ones_like(border_score), torch.zeros_like(border_score))
prediction_map = text_region.data.cpu().numpy()
border_region = border_region.data.cpu().numpy()
prediction_map[border_region==1] = 0
prob_map = probability_map.cpu().numpy()[0, 0] * 255
bord_map = border_map[0, 0].cpu().numpy() * 255
out_path = 'outputs/vis_ic15/'
image_name = data_loader.img_paths[idx].split('/')[-1].split('.')[0]
# cv2.imwrite(out_path + image_name + '_prob.png', prob_map.astype(np.uint8))
# cv2.imwrite(out_path + image_name + '_bd.png', bord_map.astype(np.uint8))
# cv2.imwrite(out_path + image_name + '_tr.png', text_region.astype(np.uint8) * 255)
# cv2.imwrite(out_path + image_name + '_fl.png', prediction_map.astype(np.uint8) * 255)
scale = (org_img.shape[1] * 1.0 / img.shape[1], org_img.shape[0] * 1.0 / img.shape[0])
bboxes = []
scale_val = scale_val.cpu().numpy()
nLabels, labels, stats, centroids = cv2.connectedComponentsWithStats(prediction_map.astype(np.uint8), connectivity=4)
t5 = time.time()
# nLabels = prediction_map.max()
# print("nLabels:", nLabels)
img_h, img_w = prediction_map.shape[:2]
for k in range(1, nLabels):
size = stats[k, cv2.CC_STAT_AREA]
# if size < 10: continue
# make segmentation map
segmap = np.zeros(prediction_map.shape, dtype=np.uint8)
segmap[labels==k] = 255
# segmap[np.logical_and(border_score > 0.7, score.cpu().numpy() < 0.05)] = 0 # remove link area
x, y = stats[k, cv2.CC_STAT_LEFT], stats[k, cv2.CC_STAT_TOP]
w, h = stats[k, cv2.CC_STAT_WIDTH], stats[k, cv2.CC_STAT_HEIGHT]
# print("xywh:", x, y, w, h, " size:", size)
niter = int(math.sqrt(size * min(w, h) / (w * h)) * 4.3)
sx, ex, sy, ey = x - niter, x + w + niter + 1, y - niter, y + h + niter + 1
# print("info:", sy, ey, sx, ex)
# boundary check
if sx < 0 : sx = 0
if sy < 0 : sy = 0
if ex >= img_w: ex = img_w
if ey >= img_h: ey = img_h
kernel = cv2.getStructuringElement(cv2.MORPH_RECT,(1 + niter, 1 + niter))
segmap[sy:ey, sx:ex] = cv2.dilate(segmap[sy:ey, sx:ex], kernel)
############### original postprocess ################
# # make segmentation map
# segmap = np.zeros(score.shape, dtype=np.uint8)
# segmap[prediction_map==k] = 255
# # contourexpand
# text_area = np.sum(segmap)
# kernel = dilated_kernel(text_area)
############## original postprocess ################
# segmap = cv2.dilate(segmap, kernel, iterations=1)
np_contours = np.roll(np.array(np.where(segmap!=0)),1,axis=0).transpose().reshape(-1,2)
rectangle = cv2.minAreaRect(np_contours)
box = cv2.boxPoints(rectangle) * 4
box = box / scale_val
box = box.astype('int32')
bboxes.append(box)
t6 = time.time()
print("infer_time:{}, post_time:{}, expand_time:{}".format(infer_time-start, post_time-infer_time, t6-t5))
# find contours
bboxes = np.array(bboxes)
num_box = bboxes.shape[0]
if args.gpus > 0:
torch.cuda.synchronize()
end = time.time()
total_frame += 1
total_time += (end - start)
sys.stdout.flush()
for bbox in bboxes:
cv2.drawContours(text_box, [bbox.reshape(4, 2)], -1, (0, 255, 0), 2)
image_name = ".".join(data_loader.img_paths[idx].split('/')[-1].split('.')[:-1])
write_result_as_txt(image_name, bboxes.reshape((-1, 8)), 'outputs/submit_ic15/')
debug(idx, data_loader.img_paths, [[text_box]], 'outputs/vis_ic15/')
def test(args):
data_loader = IC15TestLoader(long_size=args.long_size)
test_loader = torch.utils.data.DataLoader(data_loader,
batch_size=1,
shuffle=False,
num_workers=2,
drop_last=True)
submit_path = 'outputs'
if os.path.exists(submit_path):
shutil.rmtree(submit_path)
os.mkdir(submit_path)
# Setup Model
if args.arch == "resnet50":
model = models.resnet50(pretrained=True,
num_classes=7,
scale=args.scale)
elif args.arch == "resnet101":
model = models.resnet101(pretrained=True,
num_classes=7,
scale=args.scale)
elif args.arch == "resnet152":
model = models.resnet152(pretrained=True,
num_classes=7,
scale=args.scale)
for param in model.parameters():
param.requires_grad = False
model = model.cuda()
if args.resume is not None:
if os.path.isfile(args.resume):
print("Loading model and optimizer from checkpoint '{}'".format(
args.resume))
checkpoint = torch.load(args.resume)
# model.load_state_dict(checkpoint['state_dict'])
d = collections.OrderedDict()
for key, value in checkpoint['state_dict'].items():
tmp = key[7:]
d[tmp] = value
model.load_state_dict(d)
print("Loaded checkpoint '{}' (epoch {})".format(
args.resume, checkpoint['epoch']))
sys.stdout.flush()
else:
print("No checkpoint found at '{}'".format(args.resume))
sys.stdout.flush()
model.eval()
total_frame = 0.0
total_time = 0.0
for idx, (org_img, img) in enumerate(test_loader):
print('progress: %d / %d' % (idx, len(test_loader)))
sys.stdout.flush()
img = Variable(img.cuda(), volatile=True)
org_img = org_img.numpy().astype('uint8')[0]
text_box = org_img.copy()
torch.cuda.synchronize()
start = time.time()
# with torch.no_grad():
# outputs = model(img)
# outputs = torch.sigmoid(outputs)
# score = outputs[:, 0, :, :]
# outputs = outputs > args.threshold # torch.uint8
# text = outputs[:, 0, :, :]
# kernels = outputs[:, 0:args.kernel_num, :, :] * text
# score = score.squeeze(0).cpu().numpy()
# text = text.squeeze(0).cpu().numpy()
# kernels = kernels.squeeze(0).cpu().numpy()
# print(img.shape)
outputs = model(img)
score = torch.sigmoid(outputs[:, 0, :, :])
outputs = (torch.sign(outputs - args.binary_th) + 1) / 2
text = outputs[:, 0, :, :]
kernels = outputs[:, 0:args.kernel_num, :, :] * text
# print(score.shape)
# score = score.data.cpu().numpy()[0].astype(np.float32)
# text = text.data.cpu().numpy()[0].astype(np.uint8)
# kernels = kernels.data.cpu().numpy()[0].astype(np.uint8)
score = score.squeeze(0).cpu().numpy().astype(np.float32)
text = text.squeeze(0).cpu().numpy().astype(np.uint8)
kernels = kernels.squeeze(0).cpu().numpy().astype(np.uint8)
tmp_marker = kernels[-1, :, :]
# for i in range(args.kernel_num-2, -1, -1):
# sure_fg = tmp_marker
# sure_bg = kernels[i, :, :]
# watershed_source = cv2.cvtColor(sure_bg, cv2.COLOR_GRAY2BGR)
# unknown = cv2.subtract(sure_bg,sure_fg)
# ret, marker = cv2.connectedComponents(sure_fg)
# label_num = np.max(marker)
# marker += 1
# marker[unknown==1] = 0
# marker = cv2.watershed(watershed_source, marker)
# marker[marker==-1] = 1
# marker -= 1
# tmp_marker = np.asarray(marker, np.uint8)
sure_fg = kernels[-1, :, :]
sure_bg = text
watershed_source = cv2.cvtColor(sure_bg, cv2.COLOR_GRAY2BGR)
unknown = cv2.subtract(sure_bg, sure_fg)
ret, marker = cv2.connectedComponents(sure_fg)
label_num = np.max(marker)
marker += 1
marker[unknown == 1] = 0
marker = cv2.watershed(watershed_source, marker)
marker -= 1
label = marker
# label = tmp_marker
# scale = (w / marker.shape[1], h / marker.shape[0])
scale = (org_img.shape[1] * 1.0 / marker.shape[1],
org_img.shape[0] * 1.0 / marker.shape[0])
bboxes = []
# print(label_num)
for i in range(1, label_num + 1):
# get [x,y] pair, points.shape=[n, 2]
points = np.array(np.where(label == i)).transpose((1, 0))[:, ::-1]
# similar to pixellink's min_area when post-processing
if points.shape[0] < args.min_area / (args.scale * args.scale):
continue
#this filter op is very important, f-score=68.0(without) vs 69.1(with)
score_i = np.mean(score[label == i])
if score_i < args.min_score:
continue
rect = cv2.minAreaRect(points)
bbox = cv2.boxPoints(rect) * scale
# bbox=.tolist()
# bbox.append(score_i)
bbox = bbox.astype('int32')
bboxes.append(bbox.reshape(-1))
torch.cuda.synchronize()
end = time.time()
total_frame += 1
total_time += (end - start)
print('fps: %.2f' % (total_frame / total_time))
sys.stdout.flush()
for bbox in bboxes:
cv2.drawContours(text_box, [bbox.reshape(4, 2)], -1, (0, 255, 0),
2)
image_name = data_loader.img_paths[idx].split('/')[-1].split('.')[0]
write_result_as_txt(image_name, bboxes, 'outputs/submit_ic15/')
text_box = cv2.resize(text_box, (text.shape[1], text.shape[0]))
# if idx % 200 == 0:
# debug(idx, data_loader.img_paths, [[text_box]], 'outputs/vis_ic15/')
cmd = 'cd %s;zip -j %s %s/*' % ('./outputs/', 'submit_ic15.zip',
'submit_ic15')
print(cmd)
sys.stdout.flush()
util.cmd.Cmd(cmd)
import models
import torch
num_classes = 18
inputs = torch.rand([1, 3, 224, 224])
test = models.resnet34(num_classes=num_classes, pretrained='imagenet')
assert test(inputs).size()[1] == num_classes
print('ok')
test = models.resnet50(num_classes=num_classes, pretrained='imagenet')
assert test(inputs).size()[1] == num_classes
print('ok')
test = models.resnet101(num_classes=num_classes, pretrained='imagenet')
assert test(inputs).size()[1] == num_classes
print('ok')
test = models.resnet152(num_classes=num_classes, pretrained='imagenet')
assert test(inputs).size()[1] == num_classes
print('ok')
test = models.alexnet(num_classes=num_classes, pretrained='imagenet')
assert test(inputs).size()[1] == num_classes
print('ok')
test = models.densenet121(num_classes=num_classes, pretrained='imagenet')
assert test(inputs).size()[1] == num_classes
print('ok')
test = models.densenet169(num_classes=num_classes, pretrained='imagenet')
assert test(inputs).size()[1] == num_classes
print('ok')
test = models.densenet201(num_classes=num_classes, pretrained='imagenet')
assert test(inputs).size()[1] == num_classes
print('ok')
test = models.densenet201(num_classes=num_classes, pretrained='imagenet')
assert test(inputs).size()[1] == num_classes
print('ok')
def train_net(args):
torch.manual_seed(7)
np.random.seed(7)
checkpoint = args.checkpoint
start_epoch = 0
best_acc = float('-inf')
writer = SummaryWriter()
epochs_since_improvement = 0
# Initialize / load checkpoint
if checkpoint is None:
if args.network == 'r18':
model = resnet18(args)
elif args.network == 'r34':
model = resnet34(args)
elif args.network == 'r50':
model = resnet50(args)
elif args.network == 'r101':
model = resnet101(args)
elif args.network == 'r152':
model = resnet152(args)
elif args.network == 'mobile':
model = MobileNetV2()
else:
raise TypeError('network {} is not supported.'.format(
args.network))
# print(model)
model = nn.DataParallel(model)
metric_fc = ArcMarginModel(args)
metric_fc = nn.DataParallel(metric_fc)
if args.optimizer == 'sgd':
optimizer = torch.optim.SGD([{
'params': model.parameters()
}, {
'params': metric_fc.parameters()
}],
lr=args.lr,
momentum=args.mom,
weight_decay=args.weight_decay)
else:
optimizer = torch.optim.Adam([{
'params': model.parameters()
}, {
'params': metric_fc.parameters()
}],
lr=args.lr,
weight_decay=args.weight_decay)
else:
checkpoint = torch.load(checkpoint)
start_epoch = checkpoint['epoch'] + 1
epochs_since_improvement = checkpoint['epochs_since_improvement']
model = checkpoint['model']
metric_fc = checkpoint['metric_fc']
optimizer = checkpoint['optimizer']
logger = get_logger()
# Move to GPU, if available
model = model.to(device)
metric_fc = metric_fc.to(device)
# Loss function
if args.focal_loss:
criterion = FocalLoss(gamma=args.gamma).to(device)
else:
criterion = nn.CrossEntropyLoss().to(device)
# Custom dataloaders
train_dataset = ArcFaceDataset('train')
train_loader = torch.utils.data.DataLoader(train_dataset,
batch_size=args.batch_size,
shuffle=True,
num_workers=4)
scheduler = StepLR(optimizer, step_size=args.lr_step, gamma=0.1)
# Epochs
for epoch in range(start_epoch, args.end_epoch):
# One epoch's training
train_loss, train_acc = train(train_loader=train_loader,
model=model,
metric_fc=metric_fc,
criterion=criterion,
optimizer=optimizer,
epoch=epoch,
logger=logger)
writer.add_scalar('model/train_loss', train_loss, epoch)
writer.add_scalar('model/train_acc', train_acc, epoch)
# One epoch's validation
lfw_acc, threshold = lfw_test(model)
writer.add_scalar('model/valid_acc', lfw_acc, epoch)
writer.add_scalar('model/valid_thres', threshold, epoch)
# Check if there was an improvement
is_best = lfw_acc > best_acc
best_acc = max(lfw_acc, best_acc)
if not is_best:
epochs_since_improvement += 1
print("\nEpochs since last improvement: %d\n" %
(epochs_since_improvement, ))
else:
epochs_since_improvement = 0
# Save checkpoint
save_checkpoint(epoch, epochs_since_improvement, model, metric_fc,
optimizer, best_acc, is_best)
scheduler.step(epoch)
# checkpoint = 'BEST_checkpoint.tar'
# checkpoint = torch.load(checkpoint)
# model = checkpoint['model'].module
from models import resnet18, resnet34, resnet50, resnet101, resnet152, ArcMarginModel, MobileFaceNet
from utils import parse_args
from config import emb_size, PRETRAINED_DIR
args = parse_args()
if args.use_scripted_model:
model = torch.jit.load(os.path.join(PRETRAINED_DIR, args.network, 'scripted_model.pt'))
else:
if args.network == "r101":
model = resnet101(args)
model.load_state_dict(torch.load('pretrained_model/r101/insight-face-v3.pt'))
elif args.network == "mfacenet":
model = MobileFaceNet(embedding_size=emb_size)
model.load_state_dict(torch.load('pretrained_model/mfacenet/model_mobilefacenet.pth'))
model = model.to(device)
model.eval()
acc, threshold = lfw_test(model)
print('Visualizing {}...'.format(angles_file))
visualize(threshold)
print('error analysis...')
error_analysis(threshold)
def train_net(args):
torch.manual_seed(7)
np.random.seed(7)
best_loss = 100000
torch.manual_seed(7)
np.random.seed(7)
checkpoint = None
start_epoch = 0
writer = SummaryWriter()
epochs_since_improvement = 0
# Initialize / load checkpoint
if checkpoint is None:
if args.network == 'r100':
model = resnet101(args)
elif args.network == 'r50':
model = resnet50(args)
elif args.network == 'r34':
model = resnet34(args)
elif args.network == 'r18':
model = resnet18(args)
else: # 'face'
model = resnet50(args)
optimizer = torch.optim.SGD(params=filter(lambda p: p.requires_grad, model.parameters()), lr=args.lr,
momentum=args.mom, weight_decay=args.weight_decay)
else:
checkpoint = torch.load(checkpoint)
start_epoch = checkpoint['epoch'] + 1
epochs_since_improvement = checkpoint['epochs_since_improvement']
model = checkpoint['model']
optimizer = checkpoint['optimizer']
# Move to GPU, if available
model = model.to(device)
# Loss function
if args.focal_loss:
age_criterion = FocalLoss(gamma=args.gamma).to(device)
gender_criterion = FocalLoss(gamma=args.gamma).to(device)
else:
age_criterion = nn.CrossEntropyLoss().to(device)
gender_criterion = nn.CrossEntropyLoss().to(device)
criterion_info = (age_criterion, gender_criterion, args.age_weight)
# Custom dataloaders
train_dataset = AgeGenDataset('train')
train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=batch_size, shuffle=True, num_workers=workers,
pin_memory=True)
val_dataset = AgeGenDataset('valid')
val_loader = torch.utils.data.DataLoader(val_dataset, batch_size=batch_size, shuffle=False, num_workers=workers,
pin_memory=True)
scheduler = StepLR(optimizer, step_size=args.lr_step, gamma=0.1)
# Epochs
for epoch in range(start_epoch, epochs):
scheduler.step()
# One epoch's training
train_loss, train_gen_accs, train_age_mae = train(train_loader=train_loader,
model=model,
criterion_info=criterion_info,
optimizer=optimizer,
epoch=epoch)
writer.add_scalar('Train Loss', train_loss, epoch)
writer.add_scalar('Train Gender Accuracy', train_gen_accs, epoch)
writer.add_scalar('Train Age MAE', train_age_mae, epoch)
# One epoch's validation
valid_loss, valid_gen_accs, valid_age_mae = validate(val_loader=val_loader,
model=model,
criterion_info=criterion_info)
writer.add_scalar('Valid Loss', valid_loss, epoch)
writer.add_scalar('Valid Gender Accuracy', valid_gen_accs, epoch)
writer.add_scalar('Valid Age MAE', valid_age_mae, epoch)
# Check if there was an improvement
is_best = valid_loss < best_loss
best_loss = min(valid_loss, best_loss)
if not is_best:
epochs_since_improvement += 1
print("\nEpochs since last improvement: %d\n" % (epochs_since_improvement,))
else:
epochs_since_improvement = 0
# Save checkpoint
save_checkpoint(epoch, epochs_since_improvement, model, optimizer, best_loss, is_best)
def test(args):
data_loader = IC19TestLoader(long_size=args.long_size,
indic=True,
part_num=5)
test_loader = torch.utils.data.DataLoader(data_loader,
batch_size=1,
shuffle=False,
num_workers=2,
drop_last=True)
# Setup Model
if args.arch == "resnet50":
model = models.resnet50(pretrained=True,
num_classes=7,
scale=args.scale)
elif args.arch == "resnet101":
model = models.resnet101(pretrained=True,
num_classes=7,
scale=args.scale)
elif args.arch == "resnet152":
model = models.resnet152(pretrained=True,
num_classes=7,
scale=args.scale)
for param in model.parameters():
param.requires_grad = False
model = model.cuda()
if args.resume is not None:
if os.path.isfile(args.resume):
print("Loading model and optimizer from checkpoint '{}'".format(
args.resume))
checkpoint = torch.load(args.resume)
# model.load_state_dict(checkpoint['state_dict'])
d = collections.OrderedDict()
for key, value in checkpoint['state_dict'].items():
tmp = key[7:]
d[tmp] = value
model.load_state_dict(d)
print("Loaded checkpoint '{}' (epoch {})".format(
args.resume, checkpoint['epoch']))
sys.stdout.flush()
else:
print("No checkpoint found at '{}'".format(args.resume))
sys.stdout.flush()
model.eval()
total_frame = 0.0
total_time = 0.0
for idx, (org_img, img) in enumerate(test_loader):
print('progress: %d / %d' % (idx, len(test_loader)))
sys.stdout.flush()
img = Variable(img.cuda())
org_img = org_img.numpy().astype('uint8')[0]
text_box = org_img.copy()
torch.cuda.synchronize()
start = time.time()
outputs = model(img)
score = torch.sigmoid(outputs[:, 0, :, :])
outputs = (torch.sign(outputs - args.binary_th) + 1) / 2
text = outputs[:, 0, :, :]
kernels = outputs[:, 0:args.kernel_num, :, :] * text
score = score.data.cpu().numpy()[0].astype(np.float32)
text = text.data.cpu().numpy()[0].astype(np.uint8)
kernels = kernels.data.cpu().numpy()[0].astype(np.uint8)
# c++ version pse
#pred = pse(kernels, args.min_kernel_area / (args.scale * args.scale))
# python version pse
pred = pypse(kernels, args.min_kernel_area / (args.scale * args.scale))
# scale = (org_img.shape[0] * 1.0 / pred.shape[0], org_img.shape[1] * 1.0 / pred.shape[1])
scale = (org_img.shape[1] * 1.0 / pred.shape[1],
org_img.shape[0] * 1.0 / pred.shape[0])
label = pred
label_num = np.max(label) + 1
bboxes = []
for i in range(1, label_num):
points = np.array(np.where(label == i)).transpose((1, 0))[:, ::-1]
if points.shape[0] < args.min_area / (args.scale * args.scale):
continue
score_i = np.mean(score[label == i])
if score_i < args.min_score:
continue
rect = cv2.minAreaRect(points)
bbox = cv2.boxPoints(rect) * scale
bbox = bbox.astype('int32')
bboxes.append(bbox.reshape(-1))
torch.cuda.synchronize()
end = time.time()
total_frame += 1
total_time += (end - start)
print('fps: %.2f' % (total_frame / total_time))
sys.stdout.flush()
for bbox in bboxes:
cv2.drawContours(text_box, [bbox.reshape(4, 2)], -1, (0, 255, 0),
2)
image_name = data_loader.img_paths[idx].split('/')[-1].split('.')[0]
write_result_as_txt(image_name, bboxes, 'outputs/BoundingBoxCords/')
text_box = cv2.resize(text_box, (text.shape[1], text.shape[0]))
debug(idx, data_loader.img_paths, [[text_box]], 'outputs/Detections/')
import torch
import models
import os
from data_loader import BSDS_RCFLoader
from torch.utils.data import DataLoader
model = models.resnet101(pretrained=True).cuda()
init_lr = 1e-2
batch_size = 3
# resume = 'ckpt/40001.pth'
# checkpoint = torch.load(resume)
# model.load_state_dict(checkpoint)
def adjust_lr(init_lr, now_it, total_it):
power = 0.9
lr = init_lr * (1 - float(now_it) / total_it)**power
return lr
def make_optim(model, lr):
optim = torch.optim.SGD(params=model.parameters(),
lr=lr,
momentum=0.9,
weight_decay=5e-4)
return optim
def save_ckpt(model, name):
elif args.ablation == "basic":
num_input_channels = 6
elif args.ablation == "depth":
num_input_channels = 1
else:
raise NotImplementedError
logfile = open(f"{save_dir}/log_continue.txt", 'w')
def LOG(msg):
print(msg)
logfile.write(msg + '\n')
logfile.flush()
LOG('Building model...')
if args.use_count:
model = models.resnet101(num_classes=2, num_input_channels=num_input_channels, use_fc=False)
fc = FullyConnected(2048+num_categories, 2)
else:
model = models.resnet101(num_classes=2, num_input_channels=num_categories+8)
loss = nn.CrossEntropyLoss()
softmax = nn.Softmax(dim=1)
LOG('Converting to CUDA...')
model.cuda()
if args.use_count:
fc.cuda()
loss.cuda()
softmax.cuda()
LOG('Building dataset...')
def train_net(args):
torch.manual_seed(7)
np.random.seed(7)
checkpoint = args.checkpoint
start_epoch = 0
best_acc = float('-inf')
writer = SummaryWriter()
epochs_since_improvement = 0
# Initialize / load checkpoint
if checkpoint is None:
if args.network == 'r18':
model = resnet18(args)
elif args.network == 'r34':
model = resnet34(args)
elif args.network == 'r50':
model = resnet50(args)
elif args.network == 'r101':
model = resnet101(args)
elif args.network == 'r152':
model = resnet152(args)
else:
raise TypeError('network {} is not supported.'.format(
args.network))
if args.pretrained:
model.load_state_dict(torch.load('insight-face-v3.pt'))
model = nn.DataParallel(model)
metric_fc = ArcMarginModel(args)
metric_fc = nn.DataParallel(metric_fc)
if args.optimizer == 'sgd':
optimizer = torch.optim.SGD([{
'params': model.parameters()
}, {
'params': metric_fc.parameters()
}],
lr=args.lr,
momentum=args.mom,
nesterov=True,
weight_decay=args.weight_decay)
else:
optimizer = torch.optim.Adam([{
'params': model.parameters()
}, {
'params': metric_fc.parameters()
}],
lr=args.lr,
weight_decay=args.weight_decay)
else:
checkpoint = torch.load(checkpoint)
start_epoch = checkpoint['epoch'] + 1
epochs_since_improvement = checkpoint['epochs_since_improvement']
model = checkpoint['model']
metric_fc = checkpoint['metric_fc']
optimizer = checkpoint['optimizer']
# Move to GPU, if available
model = model.to(device)
metric_fc = metric_fc.to(device)
# Loss function
if args.focal_loss:
criterion = FocalLoss(gamma=args.gamma)
else:
criterion = nn.CrossEntropyLoss()
# Custom dataloaders
# train_dataset = ArcFaceDataset('train')
# train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=args.batch_size, shuffle=True,
# num_workers=num_workers)
train_dataset = ArcFaceDatasetBatched('train', img_batch_size)
train_loader = torch.utils.data.DataLoader(train_dataset,
batch_size=args.batch_size //
img_batch_size,
shuffle=True,
num_workers=num_workers,
collate_fn=batched_collate_fn)
scheduler = MultiStepLR(optimizer, milestones=[8, 16, 24, 32], gamma=0.1)
# Epochs
for epoch in range(start_epoch, args.end_epoch):
lr = optimizer.param_groups[0]['lr']
logger.info('\nCurrent effective learning rate: {}\n'.format(lr))
# print('Step num: {}\n'.format(optimizer.step_num))
writer.add_scalar('model/learning_rate', lr, epoch)
# One epoch's training
train_loss, train_top1_accs = train(train_loader=train_loader,
model=model,
metric_fc=metric_fc,
criterion=criterion,
optimizer=optimizer,
epoch=epoch)
writer.add_scalar('model/train_loss', train_loss, epoch)
writer.add_scalar('model/train_accuracy', train_top1_accs, epoch)
scheduler.step(epoch)
if args.eval_ds == "LFW":
from lfw_eval import lfw_test
# One epochs's validata
accuracy, threshold = lfw_test(model)
elif args.eval_ds == "Megaface":
from megaface_eval import megaface_test
accuracy = megaface_test(model)
else:
accuracy = -1
writer.add_scalar('model/evaluation_accuracy', accuracy, epoch)
# Check if there was an improvement
is_best = accuracy > best_acc
best_acc = max(accuracy, best_acc)
if not is_best:
epochs_since_improvement += 1
logger.info("\nEpochs since last improvement: %d\n" %
(epochs_since_improvement, ))
else:
epochs_since_improvement = 0
# Save checkpoint
save_checkpoint(epoch, epochs_since_improvement, model, metric_fc,
optimizer, best_acc, is_best, scheduler)
def train_net(args):
torch.manual_seed(7)
np.random.seed(7)
checkpoint = args.checkpoint
start_epoch = 0
best_acc = 0
writer = SummaryWriter()
epochs_since_improvement = 0
# Initialize / load checkpoint
if checkpoint is None:
if args.network == 'r18':
model = resnet18(args)
elif args.network == 'r34':
model = resnet34(args)
elif args.network == 'r50':
model = resnet50(args)
elif args.network == 'r101':
model = resnet101(args)
elif args.network == 'r152':
model = resnet152(args)
elif args.network == 'mobile':
model = MobileNet(1.0)
else:
model = resnet_face18(args.use_se)
model = nn.DataParallel(model)
metric_fc = ArcMarginModel(args)
metric_fc = nn.DataParallel(metric_fc)
if args.optimizer == 'sgd':
# optimizer = torch.optim.SGD([{'params': model.parameters()}, {'params': metric_fc.parameters()}],
# lr=args.lr, momentum=args.mom, weight_decay=args.weight_decay)
optimizer = InsightFaceOptimizer(
torch.optim.SGD([{
'params': model.parameters()
}, {
'params': metric_fc.parameters()
}],
lr=args.lr,
momentum=args.mom,
weight_decay=args.weight_decay))
else:
optimizer = torch.optim.Adam([{
'params': model.parameters()
}, {
'params': metric_fc.parameters()
}],
lr=args.lr,
weight_decay=args.weight_decay)
else:
checkpoint = torch.load(checkpoint)
start_epoch = checkpoint['epoch'] + 1
epochs_since_improvement = checkpoint['epochs_since_improvement']
model = checkpoint['model']
metric_fc = checkpoint['metric_fc']
optimizer = checkpoint['optimizer']
logger = get_logger()
# Move to GPU, if available
model = model.to(device)
metric_fc = metric_fc.to(device)
# Loss function
if args.focal_loss:
criterion = FocalLoss(gamma=args.gamma).to(device)
else:
criterion = nn.CrossEntropyLoss().to(device)
# Custom dataloaders
train_dataset = ArcFaceDataset('train')
train_loader = torch.utils.data.DataLoader(train_dataset,
batch_size=args.batch_size,
shuffle=True,
num_workers=8)
# Epochs
for epoch in range(start_epoch, args.end_epoch):
# One epoch's training
train_loss, train_top1_accs = train(train_loader=train_loader,
model=model,
metric_fc=metric_fc,
criterion=criterion,
optimizer=optimizer,
epoch=epoch,
logger=logger)
print('\nCurrent effective learning rate: {}\n'.format(optimizer.lr))
print('Step num: {}\n'.format(optimizer.step_num))
writer.add_scalar('model/train_loss', train_loss, epoch)
writer.add_scalar('model/train_accuracy', train_top1_accs, epoch)
writer.add_scalar('model/learning_rate', optimizer.lr, epoch)
# One epoch's validation
megaface_acc = megaface_test(model)
writer.add_scalar('model/megaface_accuracy', megaface_acc, epoch)
# Check if there was an improvement
is_best = megaface_acc > best_acc
best_acc = max(megaface_acc, best_acc)
if not is_best:
epochs_since_improvement += 1
print("\nEpochs since last improvement: %d\n" %
(epochs_since_improvement, ))
else:
epochs_since_improvement = 0
# Save checkpoint
save_checkpoint(epoch, epochs_since_improvement, model, metric_fc,
optimizer, best_acc, is_best)
transformed_dataset = PosterDataset(csv_file='./data.txt',
root_dir='../data/posters/posters',
transform=transforms.Compose([
Resize(),
ToTensor()
]))
train_size = int(0.8*len(transformed_dataset)+1)
test_size = int(0.2*len(transformed_dataset))
train_dataset, test_dataset = random_split(transformed_dataset, [train_size, test_size])
data_loader1 = DataLoader(train_dataset, batch_size=bs,shuffle=True)
data_loader2 = DataLoader(test_dataset, batch_size=tbs,shuffle=True)
print('train batches: ', len(data_loader1))
print('test batches: ', len(data_loader2))
device = torch.device('cuda')
model = resnet101().to(device)
criteon = nn.CrossEntropyLoss().to(device)
optimizer = optim.Adam(model.parameters(), lr=lr)
train_loss = []
train_acc = []
for epoch in range(epochs):
model.train()
losses = []
for idx, item in enumerate(data_loader1):
x, label = item['image'].to(device), item['label'].to(device)
logits = model(x)
loss = criteon(logits, label)
losses.append(loss.item())
# backporp
if __name__ == '__main__':
parser = argparse.ArgumentParser()
parser.add_argument("--weights",
type=str,
help="if specified starts from checkpoint model")
parser.add_argument("--img_size",
type=int,
default=32,
help="size of each image dimension")
parser.add_argument("--model_type",
type=str,
default="vgg",
help="vgg/resnet")
opt = parser.parse_args()
# init model
if opt.model_type == "vgg":
model = vgg11_bn().to(device)
elif opt.model_type == "resnet":
model = resnet101(sample_size=opt.img_size,
sample_duration=opt.img_size).to(device)
if opt.weights:
model.load_state_dict(torch.load(opt.weights))
else:
raise Exception("--weights is necessary!")
print(model)
evaluate(model, opt.img_size)
trainloader = torch.utils.data.DataLoader(trainset,
batch_size=BATCH_SIZE,
shuffle=True,
num_workers=4)
testloader = torch.utils.data.DataLoader(testset,
batch_size=BATCH_SIZE,
shuffle=False,
num_workers=4)
net = None
if args.depth == 18:
net = models.resnet18()
if args.depth == 50:
net = models.resnet50()
if args.depth == 101:
net = models.resnet101()
if args.depth == 152:
net = models.resnet152()
net.to(device)
criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(net.parameters(), lr=LR, weight_decay=5e-4, momentum=0.9)
if __name__ == "__main__":
best_acc = 0
for epoch in range(args.epoch):
if epoch in [60, 140, 180]:
for param_group in optimizer.param_groups:
param_group['lr'] /= 10
net.train()
sum_loss = 0.0
def test(args):
data_loader = IC15TestLoader(long_size=args.long_size)
test_loader = torch.utils.data.DataLoader(data_loader,
batch_size=1,
shuffle=False,
num_workers=2,
drop_last=True)
# Setup Model
if args.arch == "resnet50":
model = models.resnet50(pretrained=True,
num_classes=6,
scale=args.scale)
elif args.arch == "resnet101":
model = models.resnet101(pretrained=True,
num_classes=7,
scale=args.scale)
elif args.arch == "resnet152":
model = models.resnet152(pretrained=True,
num_classes=7,
scale=args.scale)
for param in model.parameters():
param.requires_grad = False
model = model.cuda()
if args.resume is not None:
if os.path.isfile(args.resume):
print("Loading model and optimizer from checkpoint '{}'".format(
args.resume))
checkpoint = torch.load(args.resume)
# model.load_state_dict(checkpoint['state_dict'])
d = collections.OrderedDict()
for key, value in checkpoint['state_dict'].items():
tmp = key[7:]
d[tmp] = value
model.load_state_dict(d)
print("Loaded checkpoint '{}' (epoch {})".format(
args.resume, checkpoint['epoch']))
sys.stdout.flush()
else:
print("No checkpoint found at '{}'".format(args.resume))
sys.stdout.flush()
model.eval()
total_frame = 0.0
total_time = 0.0
bboxs = []
bboxes = []
for idx, (org_img, img) in enumerate(test_loader):
print('progress: %d / %d' % (idx, len(test_loader)))
sys.stdout.flush()
img = Variable(img.cuda())
org_img = org_img.numpy().astype('uint8')[0]
text_box = org_img.copy()
with torch.no_grad():
outputs = model(img)
torch.cuda.synchronize()
start = time.time()
similarity_vector = outputs[0, 2:, :, :]
similarity_vector_ori = similarity_vector.permute((1, 2, 0))
score = torch.sigmoid(outputs[:, 0, :, :])
score = score.data.cpu().numpy()[0].astype(np.float32)
outputs = (torch.sign(outputs - 1.0) + 1) / 2
text = outputs[0, 0, :, :]
kernel = outputs[0, 1, :, :] * text
tag_cat, label_kernel, label_text = get_cat_tag(text, kernel)
image_name = data_loader.img_paths[idx].split('/')[-1].split('.')[0]
# cv2.imwrite('./test_result/image/text_'+image_name+'.jpg',label_text*255)
# cv2.imwrite('./test_result/image/kernel_'+image_name+'.jpg',label_kernel*255)
label_text = torch.Tensor(label_text).cuda()
label_kernel = torch.Tensor(label_kernel).cuda()
w, h, _ = similarity_vector_ori.shape
similarity_vector = similarity_vector.permute(
(1, 2, 0)).data.cpu().numpy()
bboxs = []
bboxes = []
scale = (org_img.shape[1] * 1.0 / text.shape[1],
org_img.shape[0] * 1.0 / text.shape[0])
for item in tag_cat:
similarity_vector_ori1 = similarity_vector_ori.clone()
# mask = torch.zeros((w,h)).cuda()
index_k = (label_kernel == item[0])
index_t = (label_text == item[1])
similarity_vector_k = torch.sum(
similarity_vector_ori1[index_k],
0) / similarity_vector_ori1[index_k].shape[0]
# similarity_vector_t = similarity_vector_ori1[index_t]
# similarity_vector_t = similarity_vector_ori1[index_t]
similarity_vector_ori1[~index_t] = similarity_vector_k
similarity_vector_ori1 = similarity_vector_ori1.reshape(-1, 4)
out = torch.norm((similarity_vector_ori1 - similarity_vector_k), 2,
1)
# out = torch.norm((similarity_vector_t-similarity_vector_k),2,1)
# print(out.shape)
# mask[index_t] = out
out = out.reshape(w, h)
out = out * ((text > 0).float())
# out = mask*((text>0).float())
out[out > 0.8] = 0
out[out > 0] = 1
out_im = (text * out).data.cpu().numpy()
# cv2.imwrite('./test_result/image/out_'+image_name+'.jpg',out_im*255)
points = np.array(np.where(out_im == out_im.max())).transpose(
(1, 0))[:, ::-1]
if points.shape[0] < 800:
continue
score_i = np.mean(score[out_im == out_im.max()])
if score_i < 0.93:
continue
rect = cv2.minAreaRect(points)
bbox = cv2.boxPoints(rect) * scale
bbox = bbox.astype('int32')
bboxs.append(bbox)
bboxes.append(bbox.reshape(-1))
# text_box = scale(text_box, long_size=2240)
torch.cuda.synchronize()
end = time.time()
total_frame += 1
total_time += (end - start)
print('fps: %.2f' % (total_frame / total_time))
for bbox in bboxs:
text_box = cv2.line(text_box, (bbox[0, 0], bbox[0, 1]),
(bbox[1, 0], bbox[1, 1]), (0, 0, 255), 2)
text_box = cv2.line(text_box, (bbox[1, 0], bbox[1, 1]),
(bbox[2, 0], bbox[2, 1]), (0, 0, 255), 2)
text_box = cv2.line(text_box, (bbox[2, 0], bbox[2, 1]),
(bbox[3, 0], bbox[3, 1]), (0, 0, 255), 2)
text_box = cv2.line(text_box, (bbox[3, 0], bbox[3, 1]),
(bbox[0, 0], bbox[0, 1]), (0, 0, 255), 2)
write_result_as_txt(image_name, bboxes, 'test_result/submit_ic15/')
cv2.imwrite('./test_result/image/' + image_name + '.jpg', text_box)
def main(args):
if args.checkpoint == '':
args.checkpoint = "checkpoints/ctw1500_%s_bs_%d_ep_%d" % (
args.arch, args.batch_size, args.n_epoch)
if args.pretrain:
if 'synth' in args.pretrain:
args.checkpoint += "_pretrain_synth"
else:
args.checkpoint += "_pretrain_ic17"
print(('checkpoint path: %s' % args.checkpoint))
print(('init lr: %.8f' % args.lr))
print(('schedule: ', args.schedule))
sys.stdout.flush()
if not os.path.isdir(args.checkpoint):
os.makedirs(args.checkpoint)
kernel_num = 7
min_scale = 0.4
start_epoch = 0
data_loader = CTW1500Loader(is_transform=True,
img_size=args.img_size,
kernel_num=kernel_num,
min_scale=min_scale)
train_loader = torch.utils.data.DataLoader(data_loader,
batch_size=args.batch_size,
shuffle=True,
num_workers=3,
drop_last=True,
pin_memory=True)
if args.arch == "resnet50":
model = models.resnet50(pretrained=True, num_classes=kernel_num)
elif args.arch == "resnet101":
model = models.resnet101(pretrained=True, num_classes=kernel_num)
elif args.arch == "resnet152":
model = models.resnet152(pretrained=True, num_classes=kernel_num)
model = torch.nn.DataParallel(model).cuda()
if hasattr(model.module, 'optimizer'):
optimizer = model.module.optimizer
else:
optimizer = torch.optim.SGD(model.parameters(),
lr=args.lr,
momentum=0.99,
weight_decay=5e-4)
title = 'CTW1500'
if args.pretrain:
print('Using pretrained model.')
assert os.path.isfile(
args.pretrain), 'Error: no checkpoint directory found!'
checkpoint = torch.load(args.pretrain)
model.load_state_dict(checkpoint['state_dict'])
logger = Logger(os.path.join(args.checkpoint, 'log.txt'), title=title)
logger.set_names(
['Learning Rate', 'Train Loss', 'Train Acc.', 'Train IOU.'])
elif args.resume:
print('Resuming from checkpoint.')
assert os.path.isfile(
args.resume), 'Error: no checkpoint directory found!'
checkpoint = torch.load(args.resume)
start_epoch = checkpoint['epoch']
model.load_state_dict(checkpoint['state_dict'])
optimizer.load_state_dict(checkpoint['optimizer'])
logger = Logger(os.path.join(args.checkpoint, 'log.txt'),
title=title,
resume=True)
else:
print('Training from scratch.')
logger = Logger(os.path.join(args.checkpoint, 'log.txt'), title=title)
logger.set_names(
['Learning Rate', 'Train Loss', 'Train Acc.', 'Train IOU.'])
for epoch in range(start_epoch, args.n_epoch):
adjust_learning_rate(args, optimizer, epoch)
print(('\nEpoch: [%d | %d] LR: %f' %
(epoch + 1, args.n_epoch, optimizer.param_groups[0]['lr'])))
train_loss, train_te_acc, train_ke_acc, train_te_iou, train_ke_iou = train(
train_loader, model, dice_loss, optimizer, epoch)
save_checkpoint(
{
'epoch': epoch + 1,
'state_dict': model.state_dict(),
'lr': args.lr,
'optimizer': optimizer.state_dict(),
},
checkpoint=args.checkpoint)
logger.append([
optimizer.param_groups[0]['lr'], train_loss, train_te_acc,
train_te_iou
])
logger.close()
if __name__ == "__main__":
filename = 'insight-face-v3.pt'
class HParams:
def __init__(self):
self.pretrained = False
self.use_se = True
config = HParams()
print('loading {}...'.format(filename))
start = time.time()
from models import resnet101
model = resnet101(config)
model.load_state_dict(torch.load(filename))
print('elapsed {} sec'.format(time.time() - start))
model = nn.DataParallel(model)
model = model.to(device)
model.eval()
acc, threshold = lfw_test(model)
print('Visualizing {}...'.format(angles_file))
visualize(threshold)
print('error analysis...')
error_analysis(threshold)
def main(args):
#initial setup
if args.checkpoint == '':
args.checkpoint = "checkpoints1/ic19val_%s_bs_%d_ep_%d" % (
args.arch, args.batch_size, args.n_epoch)
if args.pretrain:
if 'synth' in args.pretrain:
args.checkpoint += "_pretrain_synth"
else:
args.checkpoint += "_pretrain_ic17"
print(('checkpoint path: %s' % args.checkpoint))
print(('init lr: %.8f' % args.lr))
print(('schedule: ', args.schedule))
sys.stdout.flush()
if not os.path.isdir(args.checkpoint):
os.makedirs(args.checkpoint)
kernel_num = 7
min_scale = 0.4
start_epoch = 0
validation_split = 0.1
random_seed = 42
prev_val_loss = -1
val_loss_list = []
loggertf = tfLogger('./log/' + args.arch)
#end
#setup data loaders
data_loader = IC19Loader(is_transform=True,
img_size=args.img_size,
kernel_num=kernel_num,
min_scale=min_scale)
dataset_size = len(data_loader)
indices = list(range(dataset_size))
split = int(np.floor(validation_split * dataset_size))
np.random.seed(random_seed)
np.random.shuffle(indices)
train_incidies, val_indices = indices[split:], indices[:split]
train_sampler = SubsetRandomSampler(train_incidies)
validate_sampler = SubsetRandomSampler(val_indices)
train_loader = torch.utils.data.DataLoader(data_loader,
batch_size=args.batch_size,
num_workers=3,
drop_last=True,
pin_memory=True,
sampler=train_sampler)
validate_loader = torch.utils.data.DataLoader(data_loader,
batch_size=args.batch_size,
num_workers=3,
drop_last=True,
pin_memory=True,
sampler=validate_sampler)
#end
#Setup architecture and optimizer
if args.arch == "resnet50":
model = models.resnet50(pretrained=True, num_classes=kernel_num)
elif args.arch == "resnet101":
model = models.resnet101(pretrained=True, num_classes=kernel_num)
elif args.arch == "resnet152":
model = models.resnet152(pretrained=True, num_classes=kernel_num)
elif args.arch == "resPAnet50":
model = models.resPAnet50(pretrained=True, num_classes=kernel_num)
elif args.arch == "resPAnet101":
model = models.resPAnet101(pretrained=True, num_classes=kernel_num)
elif args.arch == "resPAnet152":
model = models.resPAnet152(pretrained=True, num_classes=kernel_num)
model = torch.nn.DataParallel(model).cuda()
if hasattr(model.module, 'optimizer'):
optimizer = model.module.optimizer
else:
optimizer = torch.optim.SGD(model.parameters(),
lr=args.lr,
momentum=0.99,
weight_decay=5e-4)
#end
#options to resume/use pretrained model/train from scratch
title = 'icdar2019MLT'
if args.pretrain:
print('Using pretrained model.')
assert os.path.isfile(
args.pretrain), 'Error: no checkpoint directory found!'
checkpoint = torch.load(args.pretrain)
model.load_state_dict(checkpoint['state_dict'])
logger = Logger(os.path.join(args.checkpoint, 'log.txt'), title=title)
logger.set_names([
'Learning Rate', 'Train Loss', 'Train Acc.', 'Train IOU.',
'Validate Loss', 'Validate Acc', 'Validate IOU'
])
elif args.resume:
print('Resuming from checkpoint.')
assert os.path.isfile(
args.resume), 'Error: no checkpoint directory found!'
checkpoint = torch.load(args.resume)
start_epoch = checkpoint['epoch']
model.load_state_dict(checkpoint['state_dict'])
optimizer.load_state_dict(checkpoint['optimizer'])
logger = Logger(os.path.join(args.checkpoint, 'log.txt'),
title=title,
resume=True)
else:
print('Training from scratch.')
logger = Logger(os.path.join(args.checkpoint, 'log.txt'), title=title)
logger.set_names([
'Learning Rate', 'Train Loss', 'Train Acc.', 'Train IOU.',
'Validate Loss', 'Validate Acc', 'Validate IOU'
])
#end
#start training model
for epoch in range(start_epoch, args.n_epoch):
adjust_learning_rate(args, optimizer, epoch)
print(('\nEpoch: [%d | %d] LR: %f' %
(epoch + 1, args.n_epoch, optimizer.param_groups[0]['lr'])))
train_loss, train_te_acc, train_ke_acc, train_te_iou, train_ke_iou = train(
train_loader, model, dice_loss, optimizer, epoch, loggertf)
val_loss, val_te_acc, val_ke_acc, val_te_iou, val_ke_iou = validate(
validate_loader, model, dice_loss)
#logging on tensorboard
loggertf.scalar_summary('Training/Accuracy', train_te_acc, epoch + 1)
loggertf.scalar_summary('Training/Loss', train_loss, epoch + 1)
loggertf.scalar_summary('Training/IoU', train_te_iou, epoch + 1)
loggertf.scalar_summary('Validation/Accuracy', val_te_acc, epoch + 1)
loggertf.scalar_summary('Validation/Loss', val_loss, epoch + 1)
loggertf.scalar_summary('Validation/IoU', val_te_iou, epoch + 1)
#end
#Boring Book Keeping
print(("End of Epoch %d", epoch + 1))
print((
"Train Loss: {loss:.4f} | Train Acc: {acc: .4f} | Train IOU: {iou_t: .4f}"
.format(loss=train_loss, acc=train_te_acc, iou_t=train_te_iou)))
print((
"Validation Loss: {loss:.4f} | Validation Acc: {acc: .4f} | Validation IOU: {iou_t: .4f}"
.format(loss=val_loss, acc=val_te_acc, iou_t=val_te_iou)))
#end
#Saving improving and Best Models
val_loss_list.append(val_loss)
if (val_loss < prev_val_loss or prev_val_loss == -1):
checkpointname = "{loss:.3f}".format(
loss=val_loss) + "_epoch" + str(epoch +
1) + "_checkpoint.pth.tar"
save_checkpoint(
{
'epoch': epoch + 1,
'state_dict': model.state_dict(),
'lr': args.lr,
'optimizer': optimizer.state_dict(),
},
checkpoint=args.checkpoint,
filename=checkpointname)
if (val_loss < min(val_loss_list)):
checkpointname = "best_checkpoint.pth.tar"
save_checkpoint(
{
'epoch': epoch + 1,
'state_dict': model.state_dict(),
'lr': args.lr,
'optimizer': optimizer.state_dict(),
},
checkpoint=args.checkpoint,
filename=checkpointname)
#end
prev_val_loss = val_loss
logger.append([
optimizer.param_groups[0]['lr'], train_loss, train_te_acc,
train_te_iou, val_loss, val_te_acc, val_te_iou
])
#end traing model
logger.close()
def train_net(args):
torch.manual_seed(7)
np.random.seed(7)
checkpoint = args.checkpoint
start_epoch = 0
best_acc = 0
writer = SummaryWriter()
epochs_since_improvement = 0
# Initialize / load checkpoint
if checkpoint is None:
if args.network == 'r18':
model = resnet18(args)
elif args.network == 'r34':
model = resnet34(args)
elif args.network == 'r50':
model = resnet50(args)
elif args.network == 'r101':
model = resnet101(args)
elif args.network == 'r152':
model = resnet152(args)
elif args.network == 'mobile':
model = MobileNet(1.0)
elif args.network == 'mr18':
print("mr18")
model = myResnet18()
else:
model = resnet_face18(args.use_se)
model = nn.DataParallel(model)
metric_fc = ArcMarginModel(args)
metric_fc = nn.DataParallel(metric_fc)
if args.optimizer == 'sgd':
optimizer = torch.optim.SGD([{
'params': model.parameters()
}, {
'params': metric_fc.parameters()
}],
lr=args.lr,
momentum=args.mom,
weight_decay=args.weight_decay)
else:
optimizer = torch.optim.Adam([{
'params': model.parameters()
}, {
'params': metric_fc.parameters()
}],
lr=args.lr,
weight_decay=args.weight_decay)
else:
checkpoint = torch.load(checkpoint)
start_epoch = checkpoint['epoch'] + 1
epochs_since_improvement = checkpoint['epochs_since_improvement']
model = checkpoint['model']
metric_fc = checkpoint['metric_fc']
optimizer = checkpoint['optimizer']
logger = get_logger()
# Move to GPU, if available
model = model.to(device)
metric_fc = metric_fc.to(device)
# Loss function
if args.focal_loss:
criterion = FocalLoss(gamma=args.gamma).to(device)
else:
criterion = nn.CrossEntropyLoss().to(device)
# Custom dataloaders
train_dataset = ArcFaceDataset('train')
train_loader = torch.utils.data.DataLoader(train_dataset,
batch_size=args.batch_size,
shuffle=True)
scheduler = StepLR(optimizer, step_size=args.lr_step, gamma=0.1)
# Epochs
for epoch in range(start_epoch, args.end_epoch):
scheduler.step()
if args.full_log:
lfw_acc, threshold = lfw_test(model)
writer.add_scalar('LFW_Accuracy', lfw_acc, epoch)
full_log(epoch)
start = datetime.now()
# One epoch's training
train_loss, train_top5_accs = train(train_loader=train_loader,
model=model,
metric_fc=metric_fc,
criterion=criterion,
optimizer=optimizer,
epoch=epoch,
logger=logger,
writer=writer)
writer.add_scalar('Train_Loss', train_loss, epoch)
writer.add_scalar('Train_Top5_Accuracy', train_top5_accs, epoch)
end = datetime.now()
delta = end - start
print('{} seconds'.format(delta.seconds))
# One epoch's validation
lfw_acc, threshold = lfw_test(model)
writer.add_scalar('LFW Accuracy', lfw_acc, epoch)
# Check if there was an improvement
is_best = lfw_acc > best_acc
best_acc = max(lfw_acc, best_acc)
if not is_best:
epochs_since_improvement += 1
print("\nEpochs since last improvement: %d\n" %
(epochs_since_improvement, ))
else:
epochs_since_improvement = 0
# Save checkpoint
save_checkpoint(epoch, epochs_since_improvement, model, metric_fc,
optimizer, best_acc, is_best)