datasets package

Subpackages

• datasets.brain package
  • Submodules
  • datasets.brain.aggregate module
  • datasets.brain.get_raw module
  • datasets.brain.input_fn module
  • datasets.brain.serialize module
  • datasets.brain.test_input_fn module
  • datasets.brain.test_serialize module
• datasets.mnist package
  • Submodules
  • datasets.mnist.get_raw module
  • datasets.mnist.input_fn module
  • datasets.mnist.serialize module
  • datasets.mnist.test_input_fn module
  • datasets.mnist.test_serialize module

Submodules

datasets.aggregate module

datasets.raw_data module

datasets.raw_data.raw_data_ingred = <ingredient_wrapper.Ingredient object>

Raw Data Ingredient

Raw data refers to data stored in some file or database format.

datasets.raw_data.raw_metacsv_ingred = <ingredient_wrapper.Ingredient object>

Raw Meta-CSV Ingredient

The meta-csv contains only the meta-information about the raw data. For example, the raw data might be a folder of medical images, and the meta-csv contains info like age, gender, subject id, image id, etc.

datasets.raw_data.raw_npz_ingred = <ingredient_wrapper.Ingredient object>

Raw NPZ Ingredient

Raw npz data refers to files stored in the numpy format npz.

datasets.serialize module

datasets.serialize.div_mod(x, y)

datasets.serialize.features_from_row(row, keys_to_handlers)

datasets.serialize.get_writer(record_pattern, split, idx, compression)

datasets.serialize.serialize_metacsv(csv_file, split_to_size)

Create TFRecords from Meta-CSV

datasets.serialize.serialize_npz(npz_file, split_to_size, xy_to_key)

Create TFRecords from NPZ

datasets.serialize.write_records(record_dir, record_pattern, keys_to_handlers, samples_per_record, split_to_size, rows_gen)

datasets.serialize.write_row(row, keys_to_handlers, writer)

datasets.serialize.write_samples(n_samples, rgen, keys_to_handlers, writer)