def initHook(obj, *file_list, **kwargs):
"""
Define the type and size of input feature and label
"""
obj.feat_size = 1024
obj.audio_size = 128
obj.label_size = 4716
obj.slots = [
DenseSlot(obj.feat_size),
DenseSlot(obj.audio_size),
SparseNonValueSlot(obj.label_size)
]
def main():
conf = parse_config("./mnist_model/trainer_config.conf.norm", "")
print conf.data_config.load_data_args
network = swig_paddle.GradientMachine.createFromConfigProto(conf.model_config)
assert isinstance(network, swig_paddle.GradientMachine) # For code hint.
network.loadParameters("./mnist_model/")
converter = DataProviderWrapperConverter(False, [DenseSlot(784)])
inArg = converter(TEST_DATA)
print network.forwardTest(inArg)
def __init__(self, train_conf, model_dir=None,
resize_dim=256, crop_dim=224,
mean_file=None,
output_layer=None,
oversample=False, is_color=True):
"""
train_conf: network configure.
model_dir: string, directory of model.
resize_dim: int, resized image size.
crop_dim: int, crop size.
mean_file: string, image mean file.
oversample: bool, oversample means multiple crops, namely five
patches (the four corner patches and the center
patch) as well as their horizontal reflections,
ten crops in all.
"""
self.train_conf = train_conf
self.model_dir = model_dir
if model_dir is None:
self.model_dir = os.path.dirname(train_conf)
self.resize_dim = resize_dim
self.crop_dims = [crop_dim, crop_dim]
self.oversample = oversample
self.is_color = is_color
self.output_layer = output_layer
if self.output_layer:
assert isinstance(self.output_layer, basestring)
self.output_layer = self.output_layer.split(",")
self.transformer = image_util.ImageTransformer(is_color = is_color)
self.transformer.set_transpose((2,0,1))
self.transformer.set_channel_swap((2,1,0))
self.mean_file = mean_file
if self.mean_file is not None:
mean = np.load(self.mean_file)['data_mean']
mean = mean.reshape(3, self.crop_dims[0], self.crop_dims[1])
self.transformer.set_mean(mean) # mean pixel
else:
# if you use three mean value, set like:
# this three mean value is calculated from ImageNet.
self.transformer.set_mean(np.array([103.939,116.779,123.68]))
conf_args = "is_test=1,use_gpu=1,is_predict=1"
conf = parse_config(train_conf, conf_args)
swig_paddle.initPaddle("--use_gpu=1")
self.network = swig_paddle.GradientMachine.createFromConfigProto(conf.model_config)
assert isinstance(self.network, swig_paddle.GradientMachine)
self.network.loadParameters(self.model_dir)
data_size = 3 * self.crop_dims[0] * self.crop_dims[1]
slots = [DenseSlot(data_size)]
is_sequence = False
self.converter = util.DataProviderWrapperConverter(is_sequence, slots)
def initHook(obj, *file_list, **kwargs):
"""
Define the type and size of input feature and label
"""
config_map = {}
configs = [args.split(':') for args in kwargs['load_data_args'].split(';')]
for confs in configs:
config_map[confs[0]] = confs[1]
logger.info(config_map)
obj.ftr_dim = int(config_map.get('ftr_dim'))
obj.label_size = int(config_map.get('label_size'))
obj.slots = [DenseSlot(obj.ftr_dim), SparseNonValueSlot(obj.label_size)]
def __init__(self,
train_conf,
use_gpu=True,
model_dir=None,
resize_dim=None,
crop_dim=None,
mean_file=None,
oversample=False,
is_color=True):
"""
train_conf: network configure.
model_dir: string, directory of model.
resize_dim: int, resized image size.
crop_dim: int, crop size.
mean_file: string, image mean file.
oversample: bool, oversample means multiple crops, namely five
patches (the four corner patches and the center
patch) as well as their horizontal reflections,
ten crops in all.
"""
self.train_conf = train_conf
self.model_dir = model_dir
if model_dir is None:
self.model_dir = os.path.dirname(train_conf)
self.resize_dim = resize_dim
self.crop_dims = [crop_dim, crop_dim]
self.oversample = oversample
self.is_color = is_color
self.transformer = image_util.ImageTransformer(is_color=is_color)
self.transformer.set_transpose((2, 0, 1))
self.mean_file = mean_file
mean = np.load(self.mean_file)['data_mean']
mean = mean.reshape(3, self.crop_dims[0], self.crop_dims[1])
self.transformer.set_mean(mean) # mean pixel
gpu = 1 if use_gpu else 0
conf_args = "is_test=1,use_gpu=%d,is_predict=1" % (gpu)
conf = parse_config(train_conf, conf_args)
swig_paddle.initPaddle("--use_gpu=%d" % (gpu))
self.network = swig_paddle.GradientMachine.createFromConfigProto(
conf.model_config)
assert isinstance(self.network, swig_paddle.GradientMachine)
self.network.loadParameters(self.model_dir)
data_size = 3 * self.crop_dims[0] * self.crop_dims[1]
slots = [DenseSlot(data_size)]
self.converter = util.DataProviderWrapperConverter(False, slots)
def initHook(obj, *file_list, **kwargs):
"""
Description: Init with a list of data file
file_list is the name list of input files.
kwargs['load_data_args'] is the value of 'load_data_args'
which can be set in config.
kwargs['load_data_args'] is organized as follows:
'dictionary path'(str)
'image feature list file'(str)
'img_feat_dim'(int)
'average norm factor for image feature'(float)
Each args is seperated by a space.
"""
str_conf_args = kwargs['load_data_args'].strip().split()
dict_file = str_conf_args[0]
img_feat_list = str_conf_args[1]
img_feat_dim = int(str_conf_args[2])
feat_avg_norm_factor = float(str_conf_args[3])
LOG.info('Dictionary path: %s', dict_file)
LOG.info('Image feature list: %s', img_feat_list)
LOG.info('Image dimension: %d', img_feat_dim)
LOG.info('Image feature norm factor: %.4f', feat_avg_norm_factor)
if os.path.isfile(dict_file):
word_dict = cPickle.load(io.open(dict_file, 'rb'))
if word_dict.get('#OOV#', -1) == -1:
word_dict['#OOV#'] = 0
if word_dict.get('$$S$$', -1) == -1:
word_dict['$$S$$'] = len(word_dict)
if word_dict.get('$$E$$', -1) == -1:
word_dict['$$E$$'] = len(word_dict)
LOG.info('Dictionary loaded with %d words', len(word_dict))
else:
LOG.fatal('Dictionary file %s does not exist!', dict_file)
sys.exit(1)
if len(file_list) == 0:
LOG.fatal('No annotation file!')
sys.exit(1)
else:
LOG.info('There are %d annotation files', len(file_list))
file_name = file_list[0].strip()
if os.path.isfile(file_name):
LOG.debug('Annotation file name: %s', file_name)
else:
LOG.fatal('Annotation file %s missing!', file_name)
sys.exit(1)
if os.path.exists(img_feat_list):
img_feat_list = io.open(img_feat_list, 'rb').readlines()
if len(img_feat_list) == 0:
LOG.fatal('No image feature file!')
sys.exit(1)
else:
LOG.info('There are %d feature files', len(img_feat_list))
else:
LOG.fatal('Image feature list %s does not exist!', img_feat_list)
sys.exit(1)
file_name = img_feat_list[0].strip()
if os.path.isfile(file_name):
LOG.debug('Image feature file name: %s', file_name)
else:
LOG.fatal('Image feature file %s missing!', file_name)
sys.exit(1)
obj.file_list = list(file_list)
obj.word_dict = word_dict
obj.features = load_image_feature(img_feat_list)
obj.img_feat_dim = img_feat_dim
obj.feat_avg_norm_factor = feat_avg_norm_factor
LOG.info('DataProvider Initialization finished')
obj.slots = [
IndexSlot(1),
DenseSlot(img_feat_dim),
IndexSlot(len(word_dict))
]