Python Pipeline.process方法代码示例

# 需要导入模块: from pipeline import Pipeline [as 别名]
# 或者: from pipeline.Pipeline import process [as 别名]
def run_batch_mode(pipeline_name, image_name):
    ext_loader = ExtensionLoader()
    pipeline = Pipeline(ext_loader.cats_container)

    pipeline_url = get_assets_resource_path_pipelines(pipeline_name)
    image_url = get_assets_resource_path_image(image_name)
    directory = os.path.join(get_batch_mode_output_path(),
                             image_name.split(".")[0] + "_" + pipeline_name.split(".")[0])

    pipeline.load_pipeline_json(pipeline_url)

    pipeline.set_input(image_url)

    if not os.path.exists(directory):
        os.makedirs(directory)

    pipeline.set_output_dir(directory)

    pipeline.process()

# 需要导入模块: from pipeline import Pipeline [as 别名]
# 或者: from pipeline.Pipeline import process [as 别名]
def main():
    # Read input file path from command line argument
    if len(sys.argv) != 2:
        print 'Please use the application as follows: python main.py <filepath>.wav'
        sys.exit(2)
    fpath = sys.argv[1]

    # If the file path ends with .wav only the given file will be processed
    # Else we assume that the input is a path to all the .wav files to be batch processed
    audio_files = []
    if fpath.endswith('.wav'):
        audio_files.append(fpath)
    else:
        audio_files.extend(glob.glob(fpath + '/*.wav'))

    # TODO: Compare human/model recognition
    # TODO: Maybe save (for instance) spectral images of sound_generation that failed

    # Setup logging
    logging.basicConfig(filename='pitch_perception.log', level=logging.INFO)

    n_channels = int(Config.get_config_option('n_channels'))

    transducer = BrianTransducer(n_channels)
    available_pitch_extractors = {'naive': NaivePitchExtractor, 'spectral': SpectralPitchExtractor(n_channels),
                                  'temporal': TemporalPitchExtractor, 'xcorr': XcorrPitchExtractor(n_channels)}

    pitch_extractor = available_pitch_extractors[Config.get_config_option('pitch_extraction')]

    # Init pipeline
    pipeline = Pipeline(transducer, pitch_extractor, test_mode=False)

    # Collect results
    results = []

    for af in audio_files:
        # Run processing
        pitch = pipeline.process(af)

        log_string = 'File: %s\tPitch: %i' % (af, pitch)
        logging.info(log_string)

        # Output final pitch
        print log_string

        results.append((af, pitch))

    # Export results
    CsvExporter.export('results.csv', pitch_extractor.__class__.__name__, results)

# 需要导入模块: from pipeline import Pipeline [as 别名]
# 或者: from pipeline.Pipeline import process [as 别名]
from tokenizer import Tokenizer
from normalizer import Normalizer
from languageTagger import LanguageTagger
from tokenTagger import CommonTwitterTokenTagger, PunctuationTagger, StopwordTagger, NamedEntityTagger
from casConsumer import DefaultWriter


input_directory = sys.argv[1]
output_directory = sys.argv[2]
corpus = sys.argv[3]

topsy_reader = TopsyReader(joinp(input_directory, "topsy"))
twitter_reader = TwitterReader(joinp(input_directory, "twitter"))
raw_reader = RawReader(joinp(input_directory, "raw"))
lang_tagger = LanguageTagger()
tokenizer = Tokenizer()
normalizer = Normalizer()
ner_tagger = NamedEntityTagger("preprocessor/config/data/multiwordlist")
stopword_tagger = StopwordTagger("preprocessor/config/data/german_stopwords")
punctuation_tagger = PunctuationTagger()
other_tagger = CommonTwitterTokenTagger()
writer = DefaultWriter(output_directory, corpus)

topsy_pipe = Pipeline(topsy_reader, lang_tagger, tokenizer, normalizer, ner_tagger, stopword_tagger, punctuation_tagger,
                      other_tagger, writer)

twitter_pipe = Pipeline(twitter_reader, lang_tagger, tokenizer, normalizer, ner_tagger, stopword_tagger,
                        punctuation_tagger, other_tagger, writer)

counter = twitter_pipe.process()
counter = topsy_pipe.process(counter=counter)