def main(self):
if self.hash != None:
t = threading.Thread(target = self.check)
t.start()
t.join()
answer = self.queue.get()
if answer != "Result failed" and answer != "Server failed":
if self.language in answer:
t = threading.Thread(target = self.download)
t.start()
t.join()
subtitles = self.queue.get()
if subtitles != "Malformed request":
imdb = Imdb(self.name)
information = imdb.main()
if information != None:
#agregar informacion a subtitulos
subtitles = "00:00:1,0 --> 00:00:20,0\nTitle: " + information.title + "\n Director:" + information.director + "\n Year:" + information.year + "\n \n" + subtitles
movie = Movie(title=information.title, director=information.director, year=information.year, hash=self.hash)
movie.save()
else:
movie = Movie(title=self.name, hash=self.hash)
movie.save()
try:
f = open("addsubs.srt",'w')
f.write(subtitles.encode("utf-8"))
f.close()
except IOError:
return None
return movie
return None
# evaluation model with pascal/voc measures
from __future__ import absolute_import, division, print_function
import os
from imdb import Imdb
from network import Network
data_dir = os.path.join(os.getcwd(), 'data')
anno_dir = os.path.join(data_dir, 'eval_annotation')
images_dir = os.path.join(data_dir, 'images')
imdb = Imdb(anno_dir, images_dir, batch_size=1)
net = Network(is_training=False)
for images, gt_boxes, gt_cls in imdb.next_batch(): # batch_size is 1
box_pred, cls_inds, scores = net.predict(images)
import time
from datetime import timedelta
from imdb import Imdb
from network import Network
data_dir = os.path.join(os.getcwd(), 'data')
anno_dir = os.path.join(data_dir, 'annotation')
images_dir = os.path.join(data_dir, 'images')
parser = argparse.ArgumentParser()
parser.add_argument('--num_epochs', type=int, default=1)
parser.add_argument('--batch_size', type=int, default=1)
parser.add_argument('--learn_rate', type=float, default=1e-3)
args = parser.parse_args()
imdb = Imdb(anno_dir, images_dir,
batch_size=args.batch_size)
net = Network(is_training=True, lr=args.learn_rate)
train_t = 0
step = 0
print('start training')
for epoch in range(1, args.num_epochs + 1):
epoch_t = time.time()
for images, gt_boxes, gt_cls in imdb.next_batch():
step, bbox_loss, iou_loss, cls_loss = net.fit(images, gt_boxes, gt_cls)
if step % 100 == 0: