Datasets:
feat: change bbox format
Browse files- visible-watermark-pita.py +13 -20
visible-watermark-pita.py
CHANGED
@@ -15,7 +15,7 @@ _VERSION = datasets.Version("1.0.0")
|
|
15 |
class WatermarkPitaConfig(datasets.BuilderConfig):
|
16 |
"""Builder Config for Food-101"""
|
17 |
|
18 |
-
def __init__(self,
|
19 |
"""BuilderConfig for Food-101.
|
20 |
|
21 |
Args:
|
@@ -28,7 +28,6 @@ class WatermarkPitaConfig(datasets.BuilderConfig):
|
|
28 |
_VERSION = datasets.Version("1.0.0")
|
29 |
|
30 |
super(WatermarkPitaConfig, self).__init__(version=_VERSION, **kwargs)
|
31 |
-
self.repository = repository
|
32 |
self.urls = urls
|
33 |
self.categories = categories
|
34 |
|
@@ -39,10 +38,19 @@ class WatermarkPita(datasets.GeneratorBasedBuilder):
|
|
39 |
BUILDER_CONFIGS = [
|
40 |
WatermarkPitaConfig(
|
41 |
name="text",
|
42 |
-
repository="data",
|
43 |
urls={"train": "data/text/train.zip", "valid": "data/text/valid.zip"},
|
44 |
categories=["text"],
|
45 |
-
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
]
|
47 |
|
48 |
_DEFAULT_CONFIG_NAME = "text"
|
@@ -66,7 +74,6 @@ class WatermarkPita(datasets.GeneratorBasedBuilder):
|
|
66 |
)
|
67 |
|
68 |
def _split_generators(self, dl_manager):
|
69 |
-
print(self.config.urls)
|
70 |
data_dir = dl_manager.download_and_extract(self.config.urls)
|
71 |
|
72 |
return [
|
@@ -86,23 +93,9 @@ class WatermarkPita(datasets.GeneratorBasedBuilder):
|
|
86 |
|
87 |
image_paths = sorted(glob(image_dir + "/*.jpg"))
|
88 |
label_paths = sorted(glob(label_dir + "/*.json"))
|
89 |
-
|
90 |
|
91 |
for idx, (image_path, label_path) in enumerate(zip(image_paths, label_paths)):
|
92 |
with open(label_path, "r") as f:
|
93 |
bbox = json.load(f)
|
94 |
|
95 |
-
objects
|
96 |
-
objects.append(
|
97 |
-
{
|
98 |
-
"label": bbox["label"],
|
99 |
-
"bbox": [
|
100 |
-
bbox["x"],
|
101 |
-
bbox["y"],
|
102 |
-
bbox["width"],
|
103 |
-
bbox["height"],
|
104 |
-
],
|
105 |
-
}
|
106 |
-
)
|
107 |
-
|
108 |
-
yield idx, {"image": image_path, "objects": objects}
|
|
|
15 |
class WatermarkPitaConfig(datasets.BuilderConfig):
|
16 |
"""Builder Config for Food-101"""
|
17 |
|
18 |
+
def __init__(self, urls, categories, **kwargs):
|
19 |
"""BuilderConfig for Food-101.
|
20 |
|
21 |
Args:
|
|
|
28 |
_VERSION = datasets.Version("1.0.0")
|
29 |
|
30 |
super(WatermarkPitaConfig, self).__init__(version=_VERSION, **kwargs)
|
|
|
31 |
self.urls = urls
|
32 |
self.categories = categories
|
33 |
|
|
|
38 |
BUILDER_CONFIGS = [
|
39 |
WatermarkPitaConfig(
|
40 |
name="text",
|
|
|
41 |
urls={"train": "data/text/train.zip", "valid": "data/text/valid.zip"},
|
42 |
categories=["text"],
|
43 |
+
),
|
44 |
+
WatermarkPitaConfig(
|
45 |
+
name="logo",
|
46 |
+
urls={"train": "data/logo/train.zip", "valid": "data/logo/valid.zip"},
|
47 |
+
categories=["logo"],
|
48 |
+
),
|
49 |
+
WatermarkPitaConfig(
|
50 |
+
name="mixed",
|
51 |
+
urls={"train": "data/mixed/train.zip", "valid": "data/mixed/valid.zip"},
|
52 |
+
categories=["logo", "text"],
|
53 |
+
),
|
54 |
]
|
55 |
|
56 |
_DEFAULT_CONFIG_NAME = "text"
|
|
|
74 |
)
|
75 |
|
76 |
def _split_generators(self, dl_manager):
|
|
|
77 |
data_dir = dl_manager.download_and_extract(self.config.urls)
|
78 |
|
79 |
return [
|
|
|
93 |
|
94 |
image_paths = sorted(glob(image_dir + "/*.jpg"))
|
95 |
label_paths = sorted(glob(label_dir + "/*.json"))
|
|
|
96 |
|
97 |
for idx, (image_path, label_path) in enumerate(zip(image_paths, label_paths)):
|
98 |
with open(label_path, "r") as f:
|
99 |
bbox = json.load(f)
|
100 |
|
101 |
+
yield idx, {"image": image_path, "objects": bbox}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|