from marshmallow import fields, validate
from ..base.schemas import BaseDatasetSchema
[docs]class MatDatasetSchema(BaseDatasetSchema):
"""
Schema for configuring a classification dataset given as mat file.
"""
mat_file = fields.String(
missing=None, description="mat file on disk", example="./data/dataset.mat",
)
mode = fields.String(
missing="train",
description="Which split to use, train or test.",
example="train",
)
csv_file = fields.String(
missing=None, description="CSV file on disk", example="./data/train.csv",
)
download = fields.Bool(
missing=False, description="Whether to download the dataset", example=True
)
[docs]class NPZDatasetSchema(BaseDatasetSchema):
"""
Schema for configuring a classification dataset given as npz file.
"""
npz_file = fields.String(
missing=None, description="npz file on disk", example="./data/dataset.npz",
)
mode = fields.String(
missing="train",
description="Which split to use, train or test.",
example="train",
)
labels = fields.List(
fields.String,
missing=None,
required=False,
description="List of labels",
)
[docs]class ClassificationDatasetSchema(BaseDatasetSchema):
"""
Schema for configuring a classification dataset.
"""
data_dir = fields.String(
missing="/", description="Dataset path on disk", example="./data/BigEarthNet/"
)
csv_file = fields.String(
missing=None, description="CSV file on disk", example="./data/train.csv",
)
[docs]class SegmentationDatasetSchema(BaseDatasetSchema):
"""
Schema for configuring a segmentation dataset.
"""
data_dir = fields.String(
missing="/", description="Dataset path on disk", example="./data/BigEarthNet/"
)
csv_file = fields.String(
missing=None, description="CSV file on disk", example="./data/train.csv",
)
[docs]class ObjectDetectionPascalDatasetSchema(BaseDatasetSchema):
"""
Schema for configuring an object detection dataset given in PASCAL VOC format.
"""
imageset_file = fields.String(
missing="/",
description="File with the image ids in the set",
example="./data/DIOR/train.txt",
)
image_dir = fields.String(
missing="/", description="Folder to the images on disk", example="./data/DIOR/"
)
annotations_dir = fields.String(
missing="/",
description="Folder with the XML annotations in VOC format",
example="./data/DIOR/Annons/",
)
[docs]class ObjectDetectionCocoDatasetSchema(BaseDatasetSchema):
"""
Schema for configuring an object detection dataset given in COCO format.
"""
data_dir = fields.String(
missing="/", description="Dataset path on disk", example="./data/DIOR/"
)
json_file = fields.String(
missing=None,
description="JSON Coco file format on disk",
example="./data/train.json",
)
hardcode_background = fields.Bool(
missing=True, description="Do we need to hardcode the background as a class?"
)
[docs]class BigEarthNetSchema(BaseDatasetSchema):
"""
Schema for configuring the BigEarthNet dataset.
"""
csv_file = fields.String(
missing=None, description="CSV file on disk", example="./data/train.csv"
)
lmdb_path = fields.String(missing=None, description="Path to the lmdb storage")
data_dir = fields.String(
missing=None, description="Dataset path on disk", example="./data/BigEarthNet/"
)
selection = fields.String(
missing="rgb", description="Read RGB channels or 13 channels", example="all/rgb"
)
version = fields.String(
missing="19 labels",
description="43 or 19 labels",
example="43 labels/19 labels",
)
import_to_lmdb = fields.Bool(
missing=False, description="Should the data be moved to LMDB"
)
bands10_mean = fields.List(
fields.Float,
missing=(429.9430203, 614.21682446, 590.23569706),
required=False,
description="List of mean values for the 3 channels",
)
bands10_std = fields.List(
fields.Float,
missing=(572.41639287, 582.87945694, 675.88746967),
required=False,
description="List of std values for the 3 channels",
)
[docs]class SpaceNet6DatasetSchema(BaseDatasetSchema):
"""
Schema for configuring the SpaceNet6 dataset.
"""
orients = fields.String(
required=False,
example="path/to/data/train/AOI_11_Roterdam/SummaryData/SAR_orientations.csv",
description="Absolute path pointing to the SAR orientations text file "
"(output of the pre-processing task",
)
root_directory = fields.String(
required=False,
example="path/to/data/train/AOI_11_Rotterdam/",
description="Root directory for the raw SpaceNet6 data set",
)
start_val_epoch = fields.Int(
required=False,
description="From which epoch should the validation period start",
)
# Train & val
folds_path = fields.String(
required=False,
example="path/to/results/folds",
description="Path to the fold csv files",
)
segmentation_directory = fields.String(
required=False,
example="path/to/results/segmentation",
description="Source directory with the target segmentation masks",
)
gt_csv = fields.String(
required=False,
description="Source file containing the ground truth segmentation data on the buildings",
)
pred_csv = fields.String(
required=False,
description="Destination file for saving the predictions from the current fold",
)
pred_folder = fields.String(
required=False,
description="Destination directory for saving the predictions from all folds",
)
edge_weight = fields.Int(
required=False, description="Weight for the building edges pixels"
)
contact_weight = fields.Int(
required=False, description="Weight for the building contact pixels"
)
# Test
test_directory = fields.String(
required=False,
example="path/to/data/train/AOI_11_Rotterdam/",
description="Root directory for the raw SpaceNet6 data set",
)
merged_pred_dir = fields.String(
required=False,
example="path/to/data/train/AOI_11_Rotterdam/",
description="Destination directory for merging the predictions from all folds",
)
solution_file = fields.String(
required=False,
example="path/to/data/results/solution.csv",
description="SpaceNet6-compliant csv destination file used for grading the challenge",
)
# Prepare
num_folds = fields.Int(
required=False, missing=10, description="Number of fold splits for the data set"
)
orients_output = fields.String(
required=False,
example="path/to/data/train/AOI_11_Roterdam/SummaryData/SAR_orientations.txt",
description="Absolute path pointing to the output SAR orientations csv file",
)
num_threads = fields.Int(
required=False,
missing=1,
description="Number of threads for parallel execution",
)
edge_width = fields.Int(
required=False,
default=3,
description="Width of the edge of buildings (in pixels)",
)
contact_width = fields.Int(
required=False,
default=9,
description="Width of the contact between (in pixels)",
)
folds_dir = fields.String(
required=False,
example="path/to/results/folds",
description="Source directory with the fold csv files",
)
[docs]class BreizhCropsSchema(BaseDatasetSchema):
"""
Schema for configuring the BreizhCrops dataset for crop type prediction.
"""
regions = fields.List(
fields.String,
required=True,
description="Brittany region (frh01..frh04)",
example="['frh01','frh01']",
)
root = fields.String(
required=True,
description="Dataset path on disk",
example="./breizhcrops_dataset",
)
year = fields.Integer(
missing=2017, description="year", validate=validate.OneOf([2017, 2018])
)
filter_length = fields.Integer(missing=0, description="filter_length")
level = fields.String(
required=True,
description="L1C or L2A",
example="L1C",
validate=validate.OneOf(["L1C", "L2A"]),
)
verbose = fields.Bool(missing=False, description="verbose") # change to true
load_timeseries = fields.Bool(missing=True, description="load_timeseries")
recompile_h5_from_csv = fields.Bool(
missing=False, description="recompile_h5_from_csv"
)
preload_ram = fields.Bool(missing=False, description="preload_ram")
[docs]class CropsDatasetSchema(BaseDatasetSchema):
"""
Schema for configuring dataset for crop type prediction.
"""
csv_file_path = fields.String(
missing=None, description="CSV file on disk", example="./data/train.csv"
)
root = fields.String(
required=True, description="Dataset path on disk", example="./slovenia-crops"
)
verbose = fields.Bool(missing=False, description="verbose")
level = fields.String(
missing="L1C",
description="L1C or L2A",
example="L1C",
validate=validate.OneOf(["L1C", "L2A"]),
)
regions = fields.List(
fields.String,
required=True,
description="Brittany region (frh01..frh04) or train/val/test",
example="['frh01','frh01']",
)
[docs]class So2SatDatasetSchema(BaseDatasetSchema):
"""
Schema for configuring the So2Sat dataset.
"""
h5_file = fields.String(
required=True, description="H5 file on disk", example="./data/train.h5"
)
