# synthtextloader = Synth80k('/home/jiachx/publicdatasets/SynthText/SynthText', target_size=768, viz=True, debug=True)
# train_loader = torch.utils.data.DataLoader(
# synthtextloader,
# batch_size=1,
# shuffle=False,
# num_workers=0,
# drop_last=True,
# pin_memory=True)
# train_batch = iter(train_loader)
# image_origin, target_gaussian_heatmap, target_gaussian_affinity_heatmap, mask = next(train_batch)
from craft import CRAFT
from torchutil import copyStateDict
net = CRAFT(freeze=True)
net.load_state_dict(
copyStateDict(torch.load('/data/CRAFT-pytorch/1-7.pth')))
net = net.cuda()
net = torch.nn.DataParallel(net)
net.eval()
dataloader = ICDAR2015(net,
'/data/CRAFT-pytorch/icdar2015',
target_size=768,
viz=True)
train_loader = torch.utils.data.DataLoader(dataloader,
batch_size=1,
shuffle=False,
num_workers=0,
drop_last=True,
pin_memory=True)
total = 0
total_sum = 0
# pin_memory=True)
# train_batch = iter(train_loader)
# image_origin, target_gaussian_heatmap, target_gaussian_affinity_heatmap, mask = next(train_batch)
from craft import CRAFT
from torchutil import copyStateDict
import argparse
parser = argparse.ArgumentParser(description='123')
parser.add_argument('--load_model',
default='',
type=str,
help='folder path to input images')
args = parser.parse_args()
net = CRAFT(freeze=True)
net.load_state_dict(copyStateDict(torch.load(args.load_model)))
net = net.cuda()
net = torch.nn.DataParallel(net)
net.eval()
dataloader = ICDAR2015(net,
'./data/icdar15/train_images/',
target_size=768,
viz=True)
train_loader = torch.utils.data.DataLoader(dataloader,
batch_size=1,
shuffle=False,
num_workers=0,
drop_last=True,
pin_memory=True)
total = 0
total_sum = 0
if __name__ == '__main__':
# synthtextloader = Synth80k('/home/jiachx/publicdatasets/SynthText/SynthText', target_size=768, viz=True, debug=True)
# train_loader = torch.utils.data.DataLoader(
# synthtextloader,
# batch_size=1,
# shuffle=False,
# num_workers=0,
# drop_last=True,
# pin_memory=True)
# train_batch = iter(train_loader)
# image_origin, target_gaussian_heatmap, target_gaussian_affinity_heatmap, mask = next(train_batch)
from detector import Detector
from torchutil import copyStateDict
net = Detector(freeze=True)
net.load_state_dict(copyStateDict(torch.load('./pretrain/1-7.pth')))
net = net.cuda()
net = torch.nn.DataParallel(net)
net.eval()
dataloader = ICDAR2015(net, './data/icdar2015', target_size=768, viz=True)
train_loader = torch.utils.data.DataLoader(dataloader,
batch_size=1,
shuffle=False,
num_workers=0,
drop_last=True,
pin_memory=True)
total = 0
total_sum = 0
for index, (opimage, region_scores, affinity_scores, confidence_mask,
confidences_mean) in enumerate(train_loader):
total += 1
if __name__ == '__main__':
# synthtextloader = Synth80k('/home/jiachx/publicdatasets/SynthText/SynthText', target_size=768, viz=True, debug=True)
# train_loader = torch.utils.data.DataLoader(
# synthtextloader,
# batch_size=1,
# shuffle=False,
# num_workers=0,
# drop_last=True,
# pin_memory=True)
# train_batch = iter(train_loader)
# image_origin, target_gaussian_heatmap, target_gaussian_affinity_heatmap, mask = next(train_batch)
from craft import CRAFT
from torchutil import copyStateDict
net = CRAFT(freeze=True)
net.load_state_dict(copyStateDict(torch.load('/ic15_iter_1300.pth')))
net = net.cuda()
net = torch.nn.DataParallel(net)
net.eval()
dataloader = ICDAR2015(net,
'/icdar2015/icdar2015train',
target_size=640,
viz=True)
train_loader = torch.utils.data.DataLoader(dataloader,
batch_size=1,
shuffle=False,
num_workers=0,
drop_last=True,
pin_memory=True)
total = 0
total_sum = 0
# image_origin, target_gaussian_heatmap, target_gaussian_affinity_heatmap, mask = next(train_batch)
from craft import CRAFT
from torchutil import copyStateDict
import argparse
parser = argparse.ArgumentParser(description='123')
parser.add_argument('--load_model', default='', type=str, help='folder path to input images')
args = parser.parse_args()
net = CRAFT(freeze=True)
net.load_state_dict(
copyStateDict(torch.load(args.load_model)))
net = net.cuda()
net = torch.nn.DataParallel(net)
net.eval()
dataloader = ICDAR2015(net, './data/icdar15/train_images/', target_size=768, viz=True)
train_loader = torch.utils.data.DataLoader(
dataloader,
batch_size=1,
shuffle=False,
num_workers=0,
drop_last=True,
pin_memory=True)
total = 0
total_sum = 0
for index, (opimage, region_scores, affinity_scores, confidence_mask, confidences_mean) in enumerate(train_loader):
total += 1
