def parse(self, url, config):
""" 解析视频信息，获取下载链接 """
CONFIG.update(config)
title = self.get_title(url)
print("[Info] 视频标题：{}".format(title))
# 创建目录结构
CONFIG["video_dir"] = touch_dir(
repair_filename(os.path.join(CONFIG['dir'], title + "-bilibili")))
if not CONFIG["video_dir"]:
print('[Error] 视频已存在！')
return []
# CONFIG["video_dir"] = touch_dir(os.path.join(CONFIG['base_dir'], "Videos"))
print("[Info] 保存路径：{}".format(
os.path.dirname(os.path.abspath(__file__)) + '\\' +
CONFIG["video_dir"]))
print("[Info] 正在解析视频信息...")
videos = self.get_videos(url)
print("[Info] 视频共有{}P".format(len(videos)))
CONFIG["videos"] = videos
for video in videos:
danmaku_path = os.path.splitext(video.path)[0] + ".xml"
self.get_danmaku(video.cid, danmaku_path)
self.get_segment(video)
return videos
# if __name__ == "__main__":
# parser = BilibiliParser()
# avid, bvid, cid, qn = [882566744, 2, 172274931, 80]
# info = parser.info_api(avid, bvid)
# parse = parser.parse_api(avid, bvid, cid, qn)
# subtitle = parser.subtitle_api(cid, avid, bvid)
# danmaku = parser.danmaku_api(cid)
# print(info, parse, subtitle, danmaku, sep='\n')
import utils
import os
import numpy as np
if __name__ == '__main__':
path = utils.get_dataset_path('UCF50')
dirs = utils.get_dirs(path)
train_list = []
val_list = []
test_list = []
train_set_percentage = 0.7
val_set_percentage_from_training = 0.2
for d in dirs:
current_dir = os.path.join(path, d)
files = utils.get_files(current_dir, '.avi', append_base=False)
files = [os.path.join(d, f) for f in files]
np.random.shuffle(files)
num_train = int(len(files) * train_set_percentage)
num_val = int(num_train * val_set_percentage_from_training)
train_list.extend(files[0:num_train - num_val])
val_list.extend(files[num_train - num_val:num_train])
test_list.extend(files[num_train:])
print(len(files), len(train_list), len(val_list), len(test_list))
save_path = path + '_lists'
utils.touch_dir(save_path)
utils.txt_write(save_path + '/trainlist.txt', train_list)
utils.txt_write(save_path + '/vallist.txt', val_list)
utils.txt_write(save_path + '/testlist.txt', test_list)
def _save_img_sequence(output_dir, img_seq, rescale=False):
utils.touch_dir(output_dir)
for i in range(img_seq.shape[0]):
_save_img(osp.join(output_dir, '%06d.png' % i), img_seq[i], rescale=rescale)
pass
def main():
np.random.seed(1)
parser = ArgumentParser()
parser.add_argument('--config', type=str, required=True, help='path of the configuration file')
parser.add_argument('--checkpoint_stage1', type=str, required=True, help='path of the stage1 checkpoint')
parser.add_argument('--checkpoint_stage2', type=str, required=True, help='path of the stage2 checkpoint')
parser.add_argument('--save_dir', type=str, required=False, help='root dir to save results', default='results/eval')
args = parser.parse_args()
config = utils.load_config(args.config)
model_config = config['model']
paths_config = config['paths']
data_dir = paths_config['data_dir']
n_points = model_config['n_pts']
n_action = model_config['n_action']
batch_size = 1
session_config = tf.ConfigProto(allow_soft_placement=True, log_device_placement=False)
session_config.gpu_options.allow_growth = True
mse_list = []
# directory setup
if not _checkpoint_exist(args.checkpoint_stage1):
raise Exception('checkpoint not found at %s' % args.checkpoint_stage1)
if not _checkpoint_exist(args.checkpoint_stage1):
raise Exception('checkpoint not found at %s' % args.checkpoint_stage2)
# start session
with tf.Session(config=session_config) as sess:
# import dataset
test_loader = SequenceDataLoader(data_dir, 'test',
n_points=n_points, n_action=n_action,
random_order=False,
randomness=False,
with_image_seq=True)
test_dataset = test_loader.get_dataset(batch_size,
repeat=False,
shuffle=False,
num_preprocess_threads=12)
# setup inputs
training_pl = tf.placeholder(tf.bool)
handle_pl = tf.placeholder(tf.string, shape=[])
base_iterator = tf.data.Iterator.from_string_handle(handle_pl, test_dataset.output_types,
test_dataset.output_shapes)
inputs = base_iterator.get_next()
# initializing models
model = FinalModel(config)
print('model initialized')
model.build(inputs)
# variables initialization
tf.logging.set_verbosity(tf.logging.INFO)
global_init = tf.global_variables_initializer()
local_init = tf.local_variables_initializer()
sess.run([global_init, local_init])
# data iterator initialization
test_iterator = test_dataset.make_initializable_iterator()
test_handle = sess.run(test_iterator.string_handle())
# checkpoint restoration
model.restore(sess, args.checkpoint_stage1)
model.restore(sess, args.checkpoint_stage2)
# iterator initialization
sess.run(test_iterator.initializer)
# running on test dataset
sample_idx = 0
n_iters = utils.training.get_n_iterations(test_loader.length(), batch_size)
feed_dict = {handle_pl: test_handle, training_pl: False}
for j in range(90):
min_mse = 1.0
for s in range(20):
outputs = model.run(sess, feed_dict)
cur_mse = np.mean(np.abs(outputs['fut_pt_raw'] - outputs['real_seq']))
if cur_mse < min_mse:
min_mse = cur_mse
# saving outputs
outputs_im = outputs['im']
outputs_real_im_seq = outputs['real_im_seq']
outputs_pred_im_seq = outputs['pred_im_seq']
# print(outputs['z'][0,:10])
outputs_mask = outputs['mask']
outputs_pred_im_crude = outputs['pred_im_crude']
outputs_current_points = outputs['current_points']
outputs_future_points = outputs['future_points']
batch_dim = outputs['im'].shape[0]
for batch_idx in range(batch_dim):
sample_save_dir = osp.join(args.save_dir, '%04d' % sample_idx)
utils.touch_dir(sample_save_dir)
_save_img(osp.join(sample_save_dir, 'input_im.png'),
outputs_im[batch_idx],
rescale=True)
_save_img(osp.join(sample_save_dir, 'current_points.png'),
outputs_current_points[batch_idx],
rescale=False)
_save_img_sequence(osp.join(sample_save_dir, 'real_seq'),
outputs_real_im_seq[batch_idx],
rescale=True)
_save_img_sequence(osp.join(sample_save_dir, 'pred_seq'),
outputs_pred_im_seq[batch_idx],
rescale=True)
_save_img_sequence(osp.join(sample_save_dir, 'mask'),
outputs_mask[batch_idx],
rescale=False)
_save_img_sequence(osp.join(sample_save_dir, 'crude'),
outputs_pred_im_crude[batch_idx],
rescale=False)
_save_img_sequence(osp.join(sample_save_dir, 'crude'),
outputs_pred_im_crude[batch_idx],
rescale=True)
_save_img_sequence(osp.join(sample_save_dir, 'pred_points'),
outputs_future_points[batch_idx],
rescale=False)
# next sample idx
pass
sample_idx += 1
print(j, min_mse)
pass
print('iteration through test set finished')
# return np.array(mse_list)
pass
print('**** Train has ', len(files_list))
save_dir = os.path.join(save_dir, 'train')
max_num_tuplex = 2000 ##200000
elif (current_subset == const.Subset.VAL.value):
files_list = utils.txt_read(
os.path.join(dataset_path + '_lists', 'vallist.txt'))
print('**** Val has ', len(files_list))
save_dir = os.path.join(save_dir, 'val')
max_num_tuplex = 100 ## 20000
elif (current_subset == const.Subset.TEST.value):
files_list = utils.txt_read(
os.path.join(dataset_path + '_lists', 'testlist.txt'))
print('*** Test has ', len(files_list))
save_dir = os.path.join(save_dir, 'test')
utils.touch_dir(save_dir)
tuple_idx = 0
tuple_idx = max(0,
utils.last_tuple_idx(save_dir) -
5) ## As caution, regenerate last 5 tuples
activity_list = sorted(utils.get_dirs(dataset_path))
print('activity_list ', len(activity_list), activity_list)
lbls_file = os.path.join(save_dir, 'lbl.pkl')
if (tuple_idx == 0):
lbls_ary = np.ones(max_num_tuplex, dtype=np.int32) * -1
## Invalid Activity
else:
lbls_ary = utils.pkl_read(lbls_file)
tuple_idx = 0
def main():
parser = ArgumentParser()
parser.add_argument('--config', type=str, required=True, help='path of the configuration file')
parser.add_argument('--checkpoint', type=str, required=True, help='path of the pretrained keypoints detector')
args = parser.parse_args()
config = utils.load_config(args.config)
paths_config = config['paths']
data_dir = paths_config['data_dir']
batch_size = 1
keypoints_root_dir = osp.join(data_dir, 'pseudo_labels')
session_config = tf.ConfigProto(allow_soft_placement=True, log_device_placement=False)
session_config.gpu_options.allow_growth = True
# directory setup
utils.touch_dir(keypoints_root_dir)
if not tf.gfile.Exists(args.checkpoint) and not tf.gfile.Exists(args.checkpoint + '.index'):
raise Exception('checkpoint not found at %s' % args.checkpoint)
# start session
with tf.Session(config=session_config) as sess:
# import dataset
train_loader = KeypointDataLoader(data_dir, 'train')
test_loader = KeypointDataLoader(data_dir, 'test')
train_dataset = train_loader.get_dataset(batch_size,
repeat=False,
shuffle=False,
num_preprocess_threads=12)
test_dataset = test_loader.get_dataset(batch_size,
repeat=False,
shuffle=False,
num_preprocess_threads=12)
# setup inputs
training_pl = tf.placeholder(tf.bool)
handle_pl = tf.placeholder(tf.string, shape=[])
base_iterator = tf.data.Iterator.from_string_handle(handle_pl, train_dataset.output_types,
train_dataset.output_shapes)
inputs = base_iterator.get_next()
# initializing models
model = KeypointModel(config)
print('model initialized')
model.build(inputs)
# variables initialization
tf.logging.set_verbosity(tf.logging.INFO)
global_init = tf.global_variables_initializer()
local_init = tf.local_variables_initializer()
sess.run([global_init, local_init])
# data iterator initialization
train_iterator = train_dataset.make_initializable_iterator()
test_iterator = test_dataset.make_initializable_iterator()
train_handle = sess.run(train_iterator.string_handle())
test_handle = sess.run(test_iterator.string_handle())
# checkpoint restoration
model.restore(sess, args.checkpoint)
# iterator initialization
sess.run(train_iterator.initializer)
sess.run(test_iterator.initializer)
# running on train dataset
n_iters = utils.training.get_n_iterations(train_loader.length(), batch_size)
feed_dict = {handle_pl: train_handle, training_pl: False}
for _ in range(n_iters):
outputs = model.run(sess, feed_dict)
_save_output(keypoints_root_dir, outputs)
pass
print('iteration through train set finished')
# running on test dataset
n_iters = utils.training.get_n_iterations(test_loader.length(), batch_size)
feed_dict = {handle_pl: test_handle, training_pl: False}
for _ in range(n_iters):
outputs = model.run(sess, feed_dict)
_save_output(keypoints_root_dir, outputs)
pass
print('iteration through test set finished')
pass