deepcell.datasets

Basic datasets can be loaded according to the following example.

from deepcell.datasets.cytoplasm import hela

# path can be modified to determine where the dataset is stored locally
(X_train,y_train),(X_test,y_test) = hela.load_data(path='hela_cytoplasm.npz',test_size=0.1)

# Details regarding dataset collection are stored in the metadata attribute
print(hela.metadata)

Tracked datasets have a dedicated load function to handle the different data structure.

from deepcell.datasets.tracked import hela

# path can be modified to determine where the dataset is stored locally
(X_train,y_train),(X_test,y_test) = hela.load_data(path='hela_tracked.npz',test_size=0.1)

deepcell.datasets.cytoplasm

Fluorescent cytoplasm datasets including the raw images and ground truth segmentation masks

deepcell.datasets.cytoplasm.a549 = <deepcell.datasets.Dataset object>
deepcell.datasets.cytoplasm.all_cells = <deepcell.datasets.Dataset object>
deepcell.datasets.cytoplasm.cho = <deepcell.datasets.Dataset object>
deepcell.datasets.cytoplasm.hela = <deepcell.datasets.Dataset object>
deepcell.datasets.cytoplasm.hela_s3 = <deepcell.datasets.Dataset object>
deepcell.datasets.cytoplasm.nih_3t3 = <deepcell.datasets.Dataset object>
deepcell.datasets.cytoplasm.pc3 = <deepcell.datasets.Dataset object>

deepcell.datasets.phase

Phase datasets including the raw images and ground truth segmentation masks

deepcell.datasets.phase.a549 = <deepcell.datasets.Dataset object>
deepcell.datasets.phase.all_cells = <deepcell.datasets.Dataset object>
deepcell.datasets.phase.cho = <deepcell.datasets.Dataset object>
deepcell.datasets.phase.hela = <deepcell.datasets.Dataset object>
deepcell.datasets.phase.hela_s3 = <deepcell.datasets.Dataset object>
deepcell.datasets.phase.nih_3t3 = <deepcell.datasets.Dataset object>
deepcell.datasets.phase.pc3 = <deepcell.datasets.Dataset object>

deepcell.datasets.tracked

Timelapse datasets of a nuclear label including the raw images and ground truth segmentation masks annotated to track cell lineages

deepcell.datasets.tracked.hek293 = <deepcell.datasets.Dataset object>
deepcell.datasets.tracked.hek293_bench = <deepcell.datasets.Dataset object>
deepcell.datasets.tracked.hela_s3 = <deepcell.datasets.Dataset object>
deepcell.datasets.tracked.hela_s3_bench = <deepcell.datasets.Dataset object>
deepcell.datasets.tracked.nih_3t3 = <deepcell.datasets.Dataset object>
deepcell.datasets.tracked.nih_3t3_bench = <deepcell.datasets.Dataset object>
deepcell.datasets.tracked.raw2647 = <deepcell.datasets.Dataset object>
deepcell.datasets.tracked.raw2647_bench = <deepcell.datasets.Dataset object>

Module contents

Builtin Datasets

class deepcell.datasets.Dataset(path, url, file_hash, metadata)[source]

Bases: object

General class for downloading datasets from S3.

Parameters

  • path (str) – path where to cache the dataset locally (relative to ~/.keras/datasets).

  • url (str) – URL of dataset in S3.

  • file_hash (str) – md5hash for checking validity of cached file.

  • metadata (dict) – miscellaneous other data for dataset

load_data(path=None, test_size=0.2, seed=0)[source]

Loads dataset.

Parameters

  • path (str) – filepath to save the data locally.

  • test_size (float) – fraction of data to reserve as test data

  • seed (int) – the seed for randomly shuffling the dataset

Returns

(x_train, y_train), (x_test, y_test).

Return type

tuple

load_tracked_data(path=None, test_size=0.2, seed=0)[source]

Loads dataset using “siamese_daughters” mode.

Parameters

  • path (str) – filepath to save the data locally.

  • test_size (float) – fraction of data to reserve as test data

  • seed (int) – the seed for randomly shuffling the dataset

Returns

(x_train, y_train), (x_test, y_test).

Return type

tuple

deepcell.datasets.hek293 = <deepcell.datasets.Dataset object>
deepcell.datasets.hela_s3 = <deepcell.datasets.Dataset object>
deepcell.datasets.mibi = <deepcell.datasets.Dataset object>
deepcell.datasets.mousebrain = <deepcell.datasets.Dataset object>
deepcell.datasets.multiplex_tissue = <deepcell.datasets.Dataset object>
deepcell.datasets.nih_3t3 = <deepcell.datasets.Dataset object>