jbloom commited on
Commit
8a3abc3
·
1 Parent(s): 21bb2fa

updated dataloader and README config section

Browse files
Files changed (2) hide show
  1. GBI-16-4D.py +51 -34
  2. README.md +8 -5
GBI-16-4D.py CHANGED
@@ -3,9 +3,12 @@ import random
3
  from glob import glob
4
  import json
5
 
 
 
6
  from astropy.io import fits
7
  import datasets
8
-
 
9
 
10
  _DESCRIPTION = (
11
  "GBI-16-4D is a dataset which is part of the AstroCompress project. It contains data "
@@ -62,53 +65,67 @@ class GBI_16_4D(datasets.GeneratorBasedBuilder):
62
  def _info(self):
63
  return datasets.DatasetInfo(
64
  description=_DESCRIPTION,
 
 
 
 
 
 
 
 
 
 
 
65
  homepage=_HOMEPAGE,
66
  license=_LICENSE,
 
67
  )
68
 
69
- def _split_generators(self, dl_manager):
70
 
71
- urls = _URLS[self.config.name]
72
-
73
- data_dir = dl_manager.download_and_extract(urls)
74
  ret = []
75
- if "train" in data_dir:
76
- ret.append(
77
- datasets.SplitGenerator(
78
- name=datasets.Split.TRAIN,
79
- gen_kwargs={"filepath": data_dir["train"],
80
- "split": "train"},
81
- ),
82
- )
83
- if "test" in data_dir:
 
 
 
 
 
 
 
 
 
 
 
 
 
 
84
  ret.append(
85
  datasets.SplitGenerator(
86
- name=datasets.Split.TEST,
87
- gen_kwargs={"filepath": data_dir["test"],
88
- "split": "test"},
 
 
89
  ),
90
  )
91
  return ret
92
 
93
- def _generate_examples(self, filepath, split, base_url=_URL):
94
  """Generate GBI-16-4D examples"""
95
 
96
- with open(filepath, "r") as f:
97
- idx = 0
98
- for line in f:
99
- task_instance_key = f"{self.config.name}-{split}-{idx}"
100
- item = json.loads(line)
101
- image_path = os.path.join(base_url, item["image"].split("./")[-1])
102
- with fits.open(image_path, memmap=False) as hdul:
103
- image_data = hdul[0].data
104
- yield task_instance_key, {"image": image_data,
105
- "ra": item["ra"],
106
- "dec": item["dec"],
107
- "pixscale": item["pixscale"],
108
- "ntimes": item["ntimes"],
109
- "nbands": item["nbands"]}
110
- idx += 1
111
-
112
 
113
  def make_split_jsonl_files(config_type="tiny", data_dir="./data",
114
  outdir="./splits", seed=42):
 
3
  from glob import glob
4
  import json
5
 
6
+
7
+ import pyarrow as pa
8
  from astropy.io import fits
9
  import datasets
10
+ from datasets import DatasetInfo, DownloadManager
11
+ from fsspec.core import url_to_fs
12
 
13
  _DESCRIPTION = (
14
  "GBI-16-4D is a dataset which is part of the AstroCompress project. It contains data "
 
65
  def _info(self):
66
  return datasets.DatasetInfo(
67
  description=_DESCRIPTION,
68
+ features=datasets.Features(
69
+ {
70
+ "image": datasets.Array4D(shape=(None, 5, 800, 800), dtype="uint16"),
71
+ "ra": datasets.Value("float64"),
72
+ "dec": datasets.Value("float64"),
73
+ "pixscale": datasets.Value("float64"),
74
+ "ntimes": datasets.Value("int64"),
75
+ "nbands": datasets.Value("int64"),
76
+ }
77
+ ),
78
+ supervised_keys=None,
79
  homepage=_HOMEPAGE,
80
  license=_LICENSE,
81
+ citation="TBD",
82
  )
83
 
84
+ def _split_generators(self, dl_manager: DownloadManager):
85
 
 
 
 
86
  ret = []
87
+ base_path = dl_manager._base_path
88
+ if base_path.startswith(datasets.config.HF_ENDPOINT):
89
+ base_path = base_path[len(datasets.config.HF_ENDPOINT):].replace("/resolve/", "@", 1)
90
+ base_path = "hf://" + base_path.lstrip("/")
91
+ _, path = url_to_fs(base_path)
92
+
93
+ for split in ["train", "test"]:
94
+ split_file_location = os.path.normpath(os.path.join(path, _URLS[self.config.name][split]))
95
+ split_file = dl_manager.download(split_file_location)
96
+ with open(split_file, encoding="utf-8") as f:
97
+ data_filenames = []
98
+ data_metadata = []
99
+ for line in f:
100
+ item = json.loads(line)
101
+ data_filenames.append(item["image"])
102
+ data_metadata.append({"ra": item["ra"],
103
+ "dec": item["dec"],
104
+ "pixscale": item["pixscale"],
105
+ "ntimes": item["ntimes"],
106
+ "nbands": item["nbands"]})
107
+
108
+ data_urls = [os.path.normpath(os.path.join(path,data_filename)) for data_filename in data_filenames]
109
+ data_files = [dl_manager.download(data_url) for data_url in data_urls]
110
  ret.append(
111
  datasets.SplitGenerator(
112
+ name=datasets.Split.TRAIN if split == "train" else datasets.Split.TEST,
113
+ gen_kwargs={"filepaths": data_files,
114
+ "split_file": split_file,
115
+ "split": split,
116
+ "data_metadata": data_metadata},
117
  ),
118
  )
119
  return ret
120
 
121
+ def _generate_examples(self, filepaths, split_file, split, data_metadata):
122
  """Generate GBI-16-4D examples"""
123
 
124
+ for idx, (filepath, item) in enumerate(zip(filepaths, data_metadata)):
125
+ task_instance_key = f"{self.config.name}-{split}-{idx}"
126
+ with fits.open(filepath, memmap=False) as hdul:
127
+ image_data = hdul[0].data.tolist()
128
+ yield task_instance_key, {**{"image": image_data}, **item}
 
 
 
 
 
 
 
 
 
 
 
129
 
130
  def make_split_jsonl_files(config_type="tiny", data_dir="./data",
131
  outdir="./splits", seed=42):
README.md CHANGED
@@ -9,10 +9,13 @@ dataset_info:
9
  config_name: tiny
10
  features:
11
  - name: image
12
- sequence:
13
- sequence:
14
- sequence:
15
- sequence: uint16
 
 
 
16
  - name: ra
17
  dtype: float64
18
  - name: dec
@@ -30,7 +33,7 @@ dataset_info:
30
  - name: test
31
  num_bytes: 352881364
32
  num_examples: 1
33
- download_size: 692
34
  dataset_size: 911075540
35
  ---
36
  GBI-16-4D Dataset
 
9
  config_name: tiny
10
  features:
11
  - name: image
12
+ dtype:
13
+ array4_d:
14
+ shape:
15
+ - 5
16
+ - 800
17
+ - 800
18
+ dtype: uint16
19
  - name: ra
20
  dtype: float64
21
  - name: dec
 
33
  - name: test
34
  num_bytes: 352881364
35
  num_examples: 1
36
+ download_size: 908845172
37
  dataset_size: 911075540
38
  ---
39
  GBI-16-4D Dataset