Chris Oswald
commited on
Commit
•
61fd3fa
1
Parent(s):
b64741a
initial commit
Browse files
SPIDER.py
ADDED
@@ -0,0 +1,236 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
|
2 |
+
#
|
3 |
+
# Licensed under the Apache License, Version 2.0 (the "License");
|
4 |
+
# you may not use this file except in compliance with the License.
|
5 |
+
# You may obtain a copy of the License at
|
6 |
+
#
|
7 |
+
# http://www.apache.org/licenses/LICENSE-2.0
|
8 |
+
#
|
9 |
+
# Unless required by applicable law or agreed to in writing, software
|
10 |
+
# distributed under the License is distributed on an "AS IS" BASIS,
|
11 |
+
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
12 |
+
# See the License for the specific language governing permissions and
|
13 |
+
# limitations under the License.
|
14 |
+
# TODO: Address all TODOs and remove all explanatory comments
|
15 |
+
"""TODO: Add a description here."""
|
16 |
+
|
17 |
+
|
18 |
+
import csv
|
19 |
+
import json
|
20 |
+
import os
|
21 |
+
from typing import Dict, List, Optional, Set
|
22 |
+
|
23 |
+
import numpy as np
|
24 |
+
|
25 |
+
import datasets
|
26 |
+
|
27 |
+
|
28 |
+
# TODO: Add BibTeX citation
|
29 |
+
# Find for instance the citation on arxiv or on the dataset repo/website
|
30 |
+
_CITATION = """\
|
31 |
+
@InProceedings{huggingface:dataset,
|
32 |
+
title = {A great new dataset},
|
33 |
+
author={huggingface, Inc.
|
34 |
+
},
|
35 |
+
year={2020}
|
36 |
+
}
|
37 |
+
"""
|
38 |
+
|
39 |
+
# TODO: Add description of the dataset here
|
40 |
+
# You can copy an official description
|
41 |
+
_DESCRIPTION = """\
|
42 |
+
This new dataset is designed to solve this great NLP task and is crafted with a lot of care.
|
43 |
+
"""
|
44 |
+
|
45 |
+
_HOMEPAGE = "https://zenodo.org/records/10159290"
|
46 |
+
|
47 |
+
_LICENSE = """Creative Commons Attribution 4.0 International License \
|
48 |
+
(https://creativecommons.org/licenses/by/4.0/legalcode)"""
|
49 |
+
|
50 |
+
# TODO: Add link to the official dataset URLs here
|
51 |
+
# The HuggingFace Datasets library doesn't host the datasets but only points to the original files.
|
52 |
+
# This can be an arbitrary nested dict/list of URLs (see below in `_split_generators` method)
|
53 |
+
_URLS = {
|
54 |
+
"first_domain": {
|
55 |
+
"images":"https://zenodo.org/records/10159290/files/images.zip",
|
56 |
+
"masks":"https://zenodo.org/records/10159290/files/masks.zip",
|
57 |
+
"overview":"https://zenodo.org/records/10159290/files/overview.csv",
|
58 |
+
"gradings":"https://zenodo.org/records/10159290/files/radiological_gradings.csv",
|
59 |
+
}
|
60 |
+
}
|
61 |
+
|
62 |
+
class SPIDER(datasets.GeneratorBasedBuilder):
|
63 |
+
"""TODO: Short description of my dataset."""
|
64 |
+
|
65 |
+
VERSION = datasets.Version("1.1.0")
|
66 |
+
|
67 |
+
# This is an example of a dataset with multiple configurations.
|
68 |
+
# If you don't want/need to define several sub-sets in your dataset,
|
69 |
+
# just remove the BUILDER_CONFIG_CLASS and the BUILDER_CONFIGS attributes.
|
70 |
+
|
71 |
+
# If you need to make complex sub-parts in the datasets with configurable options
|
72 |
+
# You can create your own builder configuration class to store attribute, inheriting from datasets.BuilderConfig
|
73 |
+
# BUILDER_CONFIG_CLASS = MyBuilderConfig
|
74 |
+
|
75 |
+
# You will be able to load one or the other configurations in the following list with
|
76 |
+
# data = datasets.load_dataset('my_dataset', 'first_domain')
|
77 |
+
# data = datasets.load_dataset('my_dataset', 'second_domain')
|
78 |
+
BUILDER_CONFIGS = [
|
79 |
+
datasets.BuilderConfig(name="first_domain", version=VERSION, description="This part of my dataset covers a first domain"),
|
80 |
+
datasets.BuilderConfig(name="second_domain", version=VERSION, description="This part of my dataset covers a second domain"),
|
81 |
+
]
|
82 |
+
|
83 |
+
DEFAULT_CONFIG_NAME = "first_domain" # It's not mandatory to have a default configuration. Just use one if it make sense.
|
84 |
+
|
85 |
+
def _info(self):
|
86 |
+
# TODO: This method specifies the datasets.DatasetInfo object which contains informations and typings for the dataset
|
87 |
+
if self.config.name == "first_domain": # This is the name of the configuration selected in BUILDER_CONFIGS above
|
88 |
+
features = datasets.Features(
|
89 |
+
{
|
90 |
+
"sentence": datasets.Value("string"),
|
91 |
+
"option1": datasets.Value("string"),
|
92 |
+
"answer": datasets.Value("string")
|
93 |
+
# These are the features of your dataset like images, labels ...
|
94 |
+
}
|
95 |
+
)
|
96 |
+
else: # This is an example to show how to have different features for "first_domain" and "second_domain"
|
97 |
+
features = datasets.Features(
|
98 |
+
{
|
99 |
+
"sentence": datasets.Value("string"),
|
100 |
+
"option2": datasets.Value("string"),
|
101 |
+
"second_domain_answer": datasets.Value("string")
|
102 |
+
# These are the features of your dataset like images, labels ...
|
103 |
+
}
|
104 |
+
)
|
105 |
+
return datasets.DatasetInfo(
|
106 |
+
# This is the description that will appear on the datasets page.
|
107 |
+
description=_DESCRIPTION,
|
108 |
+
# This defines the different columns of the dataset and their types
|
109 |
+
features=features, # Here we define them above because they are different between the two configurations
|
110 |
+
# If there's a common (input, target) tuple from the features, uncomment supervised_keys line below and
|
111 |
+
# specify them. They'll be used if as_supervised=True in builder.as_dataset.
|
112 |
+
# supervised_keys=("sentence", "label"),
|
113 |
+
# Homepage of the dataset for documentation
|
114 |
+
homepage=_HOMEPAGE,
|
115 |
+
# License for the dataset if available
|
116 |
+
license=_LICENSE,
|
117 |
+
# Citation for the dataset
|
118 |
+
citation=_CITATION,
|
119 |
+
)
|
120 |
+
|
121 |
+
def _split_generators(self, dl_manager):
|
122 |
+
# TODO: This method is tasked with downloading/extracting the data and defining the splits depending on the configuration
|
123 |
+
# If several configurations are possible (listed in BUILDER_CONFIGS), the configuration selected by the user is in self.config.name
|
124 |
+
|
125 |
+
# dl_manager is a datasets.download.DownloadManager that can be used to download and extract URLS
|
126 |
+
# It can accept any type or nested list/dict and will give back the same structure with the url replaced with path to local files.
|
127 |
+
# By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
|
128 |
+
urls = _URLS[self.config.name]
|
129 |
+
paths_dict = dl_manager.download_and_extract(urls)
|
130 |
+
|
131 |
+
return [
|
132 |
+
datasets.SplitGenerator(
|
133 |
+
name=datasets.Split.TRAIN,
|
134 |
+
# These kwargs will be passed to _generate_examples
|
135 |
+
gen_kwargs={
|
136 |
+
"paths_dict": paths_dict,
|
137 |
+
"split": "train",
|
138 |
+
},
|
139 |
+
),
|
140 |
+
datasets.SplitGenerator(
|
141 |
+
name=datasets.Split.VALIDATION,
|
142 |
+
# These kwargs will be passed to _generate_examples
|
143 |
+
gen_kwargs={
|
144 |
+
"paths_dict": paths_dict,
|
145 |
+
"split": "dev",
|
146 |
+
},
|
147 |
+
),
|
148 |
+
datasets.SplitGenerator(
|
149 |
+
name=datasets.Split.TEST,
|
150 |
+
# These kwargs will be passed to _generate_examples
|
151 |
+
gen_kwargs={
|
152 |
+
"paths_dict": paths_dict,
|
153 |
+
"split": "test"
|
154 |
+
},
|
155 |
+
),
|
156 |
+
]
|
157 |
+
|
158 |
+
# method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
|
159 |
+
def _generate_examples(self, paths_dict, split):
|
160 |
+
# TODO: This method handles input defined in _split_generators to yield (key, example) tuples from the dataset.
|
161 |
+
# The `key` is for legacy reasons (tfds) and is not important in itself, but must be unique for each example.
|
162 |
+
|
163 |
+
# Generate train/validate/test partitions of patient IDs
|
164 |
+
np.random.seed(9999)
|
165 |
+
N_PATIENTS = 257 #TODO: make hardcoded values dynamic
|
166 |
+
VALIDATE_SHARE = 0.3
|
167 |
+
TEST_SHARE = 0.2
|
168 |
+
TRAIN_SHARE = (1.0 - VALIDATE_SHARE - TEST_SHARE)
|
169 |
+
partition = np.random.choice(
|
170 |
+
['train', 'dev', 'test'],
|
171 |
+
p=[TRAIN_SHARE, VALIDATE_SHARE, TEST_SHARE],
|
172 |
+
size=N_PATIENTS,
|
173 |
+
)
|
174 |
+
patient_ids = (np.arange(N_PATIENTS) + 1)
|
175 |
+
train_ids = set(patient_ids[partition == 'train'])
|
176 |
+
validate_ids = set(patient_ids[partition == 'dev'])
|
177 |
+
test_ids = set(patient_ids[partition == 'test'])
|
178 |
+
assert len(train_ids.union(validate_ids, test_ids)) == N_PATIENTS
|
179 |
+
|
180 |
+
|
181 |
+
|
182 |
+
# Import patient/scanner data and radiological gradings data
|
183 |
+
overview_data = import_csv_data(paths_dict['overview'])
|
184 |
+
grades_data = import_csv_data(paths_dict['gradings'])
|
185 |
+
|
186 |
+
# Import image and mask data
|
187 |
+
image_files = [
|
188 |
+
file for file in os.listdir(os.path.join(paths_dict['images'], 'images'))
|
189 |
+
if file.endswith('.mha')
|
190 |
+
]
|
191 |
+
assert len(image_files) > 0, "No image files found--check directory path."
|
192 |
+
|
193 |
+
mask_files = [
|
194 |
+
file for file in os.listdir(os.path.join(paths_dict['masks'], 'masks'))
|
195 |
+
if file.endswith('.mha')
|
196 |
+
]
|
197 |
+
assert len(mask_files) > 0, "No mask files found--check directory path."
|
198 |
+
|
199 |
+
images = []
|
200 |
+
masks = []
|
201 |
+
if split == 'train':
|
202 |
+
for patient_id in train_ids:
|
203 |
+
|
204 |
+
|
205 |
+
elif split == 'validate':
|
206 |
+
|
207 |
+
elif split == 'test':
|
208 |
+
|
209 |
+
|
210 |
+
def import_csv_data(filepath: str) -> List[Dict[str, str]]:
|
211 |
+
"""Import all rows of CSV file."""
|
212 |
+
results = []
|
213 |
+
with open(filepath, encoding='utf-8') as f:
|
214 |
+
reader = csv.DictReader(f)
|
215 |
+
for line in reader:
|
216 |
+
results.append(line)
|
217 |
+
return results
|
218 |
+
|
219 |
+
|
220 |
+
|
221 |
+
with open(filepath, encoding="utf-8") as f:
|
222 |
+
for key, row in enumerate(f):
|
223 |
+
data = json.loads(row)
|
224 |
+
if self.config.name == "first_domain":
|
225 |
+
# Yields examples as (key, example) tuples
|
226 |
+
yield key, {
|
227 |
+
"sentence": data["sentence"],
|
228 |
+
"option1": data["option1"],
|
229 |
+
"answer": "" if split == "test" else data["answer"],
|
230 |
+
}
|
231 |
+
else:
|
232 |
+
yield key, {
|
233 |
+
"sentence": data["sentence"],
|
234 |
+
"option2": data["option2"],
|
235 |
+
"second_domain_answer": "" if split == "test" else data["second_domain_answer"],
|
236 |
+
}
|