Thinh Huynh Nguyen Truong commited on
Commit
6896071
1 Parent(s): cf18101
Files changed (2) hide show
  1. requirements.txt +2 -0
  2. test.py +47 -35
requirements.txt CHANGED
@@ -1 +1,3 @@
 
1
  datasets
 
 
1
+ # Python 3.10.6
2
  datasets
3
+ Pillow
test.py CHANGED
@@ -18,6 +18,8 @@
18
  import csv
19
  import json
20
  import os
 
 
21
 
22
  import datasets
23
  from datasets import DownloadManager
@@ -56,6 +58,19 @@ _URLS = {
56
  _BASE_URL = ""
57
 
58
 
 
 
 
 
 
 
 
 
 
 
 
 
 
59
  # TODO: Name of the dataset usually matches the script name with CamelCase instead of snake_case
60
  class Test(datasets.GeneratorBasedBuilder):
61
  """TODO: Short description of my dataset."""
@@ -98,6 +113,7 @@ class Test(datasets.GeneratorBasedBuilder):
98
  "depth": datasets.Image(),
99
  "rgb": datasets.Image(),
100
  "gt": datasets.Image(),
 
101
  # These are the features of your dataset like images, labels ...
102
  }
103
  ), # Here we define them above because they are different between the two configurations
@@ -121,52 +137,48 @@ class Test(datasets.GeneratorBasedBuilder):
121
  # By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
122
  urls = _URLS[self.config.name]
123
  data_dir = dl_manager.download_and_extract(urls)
 
 
 
 
 
 
 
 
 
124
  return [
125
  datasets.SplitGenerator(
126
  name=datasets.Split.TRAIN,
127
  # These kwargs will be passed to _generate_examples
128
- gen_kwargs={
129
- "filepath": os.path.join(data_dir, "train.jsonl"),
130
- "split": "train",
131
- },
132
  ),
133
  datasets.SplitGenerator(
134
  name=datasets.Split.VALIDATION,
135
  # These kwargs will be passed to _generate_examples
136
- gen_kwargs={
137
- "filepath": os.path.join(data_dir, "dev.jsonl"),
138
- "split": "dev",
139
- },
140
- ),
141
- datasets.SplitGenerator(
142
- name=datasets.Split.TEST,
143
- # These kwargs will be passed to _generate_examples
144
- gen_kwargs={
145
- "filepath": os.path.join(data_dir, "test.jsonl"),
146
- "split": "test",
147
- },
148
  ),
 
 
 
 
 
149
  ]
150
 
151
  # method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
152
- def _generate_examples(self, filepath: str, split: str):
153
  # TODO: This method handles input defined in _split_generators to yield (key, example) tuples from the dataset.
154
  # The `key` is for legacy reasons (tfds) and is not important in itself, but must be unique for each example.
155
- with open(filepath, encoding="utf-8") as f:
156
- for key, row in enumerate(f):
157
- data = json.loads(row)
158
- if self.config.name == "first_domain":
159
- # Yields examples as (key, example) tuples
160
- yield key, {
161
- "sentence": data["sentence"],
162
- "option1": data["option1"],
163
- "answer": "" if split == "test" else data["answer"],
164
- }
165
- else:
166
- yield key, {
167
- "sentence": data["sentence"],
168
- "option2": data["option2"],
169
- "second_domain_answer": ""
170
- if split == "test"
171
- else data["second_domain_answer"],
172
- }
 
18
  import csv
19
  import json
20
  import os
21
+ from typing import Dict, List
22
+ from PIL import Image
23
 
24
  import datasets
25
  from datasets import DownloadManager
 
58
  _BASE_URL = ""
59
 
60
 
61
+ def get_download_url(config_name: str, partition: str) -> str:
62
+ """Get download URL based on config name and parition (train/dev/test)
63
+
64
+ Args:
65
+ config_name (str): can be "v1", "v2",...
66
+ partition (str): can be "train", "dev" or "test"
67
+
68
+ Returns:
69
+ str: URL to download file
70
+ """
71
+ return f"https://huggingface.co/datasets/RGBD-SOD/test/tree/main/data{config_name}/{partition}.zip"
72
+
73
+
74
  # TODO: Name of the dataset usually matches the script name with CamelCase instead of snake_case
75
  class Test(datasets.GeneratorBasedBuilder):
76
  """TODO: Short description of my dataset."""
 
113
  "depth": datasets.Image(),
114
  "rgb": datasets.Image(),
115
  "gt": datasets.Image(),
116
+ "name": datasets.Value("string"),
117
  # These are the features of your dataset like images, labels ...
118
  }
119
  ), # Here we define them above because they are different between the two configurations
 
137
  # By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
138
  urls = _URLS[self.config.name]
139
  data_dir = dl_manager.download_and_extract(urls)
140
+ train_dir = dl_manager.download_and_extract(
141
+ get_download_url(self.config.name, "train")
142
+ )
143
+ dev_dir = dl_manager.download_and_extract(
144
+ get_download_url(self.config.name, "dev")
145
+ )
146
+ # test_dir = dl_manager.download_and_extract(
147
+ # get_download_url(self.config.name, "test")
148
+ # )
149
  return [
150
  datasets.SplitGenerator(
151
  name=datasets.Split.TRAIN,
152
  # These kwargs will be passed to _generate_examples
153
+ gen_kwargs={"dir_path": train_dir},
 
 
 
154
  ),
155
  datasets.SplitGenerator(
156
  name=datasets.Split.VALIDATION,
157
  # These kwargs will be passed to _generate_examples
158
+ gen_kwargs={"dir_path": dev_dir},
 
 
 
 
 
 
 
 
 
 
 
159
  ),
160
+ # datasets.SplitGenerator(
161
+ # name=datasets.Split.TEST,
162
+ # # These kwargs will be passed to _generate_examples
163
+ # gen_kwargs={"dir_path": test_dir},
164
+ # ),
165
  ]
166
 
167
  # method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
168
+ def _generate_examples(self, dir_path: str):
169
  # TODO: This method handles input defined in _split_generators to yield (key, example) tuples from the dataset.
170
  # The `key` is for legacy reasons (tfds) and is not important in itself, but must be unique for each example.
171
+ with open(os.path.join(dir_path, "metadata.json"), "r") as f:
172
+ json_object = json.load(f)
173
+
174
+ metadata: List[Dict[str, str]] = json_object["metadata"]
175
+
176
+ for key, row in enumerate(metadata):
177
+ yield key, {
178
+ "name": row["name"],
179
+ "rgb": Image.open(os.path.join(dir_path, metadata["rgb"]), mode="RGB"),
180
+ "gt": Image.open(os.path.join(dir_path, metadata["gt"]), mode="L"),
181
+ "depth": Image.open(
182
+ os.path.join(dir_path, metadata["depth"]), mode="L"
183
+ ),
184
+ }