add custom script
Browse files- cell_benchmark.py +87 -0
cell_benchmark.py
ADDED
@@ -0,0 +1,87 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import datasets
|
3 |
+
#from datasets import DownloadManager, DatasetInfo
|
4 |
+
|
5 |
+
_DESCRIPTION = """\
|
6 |
+
A segmentation dataset for [TODO: complete...]
|
7 |
+
"""
|
8 |
+
|
9 |
+
|
10 |
+
_HOMEPAGE = "https://huggingface.co/datasets/alkzar90/cell_benchmark"
|
11 |
+
_EXTENSION = [".jpg", ".png"]
|
12 |
+
_URL_BASE = "https://huggingface.co/datasets/alkzar90/cell_benchmark/resolve/main/data/"
|
13 |
+
_SPLIT_URLS = {
|
14 |
+
"train": _URL_BASE + "train.zip",
|
15 |
+
"val": _URL_BASE + "val.zip",
|
16 |
+
"test": _URL_BASE + "test.zip",
|
17 |
+
"masks": _URL_BASE + "masks.zip",
|
18 |
+
}
|
19 |
+
|
20 |
+
|
21 |
+
|
22 |
+
class Cellsegmentation(datasets.GeneratorBasedBuilder):
|
23 |
+
|
24 |
+
def _info(self) -> DatasetInfo:
|
25 |
+
features = datasets.Features({
|
26 |
+
"image": datasets.Image(),
|
27 |
+
"masks": datasets.Image(),
|
28 |
+
"path" : datasets.Value("string"),
|
29 |
+
})
|
30 |
+
return datasets.DatasetInfo(
|
31 |
+
description=_DESCRIPTION,
|
32 |
+
features=datasets.Features(features),
|
33 |
+
supervised_keys=("image", "masks"),
|
34 |
+
homepage_HOMEPAGE,
|
35 |
+
citation="",
|
36 |
+
)
|
37 |
+
|
38 |
+
|
39 |
+
def _split_generators(self, dl_manager):
|
40 |
+
data_files = dl_manager.download_and_extract(_SPLIT_URLS)
|
41 |
+
splits = [
|
42 |
+
datasets.SplitGenerator(
|
43 |
+
name=datasets.Split.TRAIN,
|
44 |
+
gen_kwargs={
|
45 |
+
"files" : dl_manager.iter_files([data_files["train"]]),
|
46 |
+
"masks": data_files["masks"],
|
47 |
+
"split": "training",
|
48 |
+
},
|
49 |
+
),
|
50 |
+
datasets.SplitGenerator(
|
51 |
+
name=datasets.Split.VALIDATION,
|
52 |
+
gen_kwargs={
|
53 |
+
"files" : dl_manager.iter_files([data_files["validation"]]),
|
54 |
+
"masks": data_files["masks"],
|
55 |
+
"split": "validation",
|
56 |
+
},
|
57 |
+
),
|
58 |
+
datasets.SplitGenerator(
|
59 |
+
name=datasets.Split.TEST,
|
60 |
+
gen_kwargs={
|
61 |
+
"files" : dl_manager.iter_files([data_files["test"]]),
|
62 |
+
"masks": data_files["masks"],
|
63 |
+
"split": "test",
|
64 |
+
}
|
65 |
+
)
|
66 |
+
]
|
67 |
+
return splits
|
68 |
+
|
69 |
+
|
70 |
+
def _generate_examples(self, files, masks, split):
|
71 |
+
mask_path = os.path.basename(masks)
|
72 |
+
for i, path in enumereate(files):
|
73 |
+
file_name = os.path.basename(path)
|
74 |
+
yield i, {
|
75 |
+
"image": path,
|
76 |
+
"masks": mask_path + "mask_" + file_name.replace("jpg", "png"),
|
77 |
+
"path": path,
|
78 |
+
}
|
79 |
+
|
80 |
+
|
81 |
+
|
82 |
+
|
83 |
+
|
84 |
+
|
85 |
+
|
86 |
+
|
87 |
+
|