fuliucansheng
commited on
Commit
·
2b5cf97
1
Parent(s):
e355620
add dataset minicoco
Browse files- .gitattributes +1 -0
- minicoco.py +531 -0
- minicoco.tar.gz +3 -0
.gitattributes
CHANGED
@@ -14,3 +14,4 @@
|
|
14 |
*.pb filter=lfs diff=lfs merge=lfs -text
|
15 |
*.pt filter=lfs diff=lfs merge=lfs -text
|
16 |
*.pth filter=lfs diff=lfs merge=lfs -text
|
|
|
|
14 |
*.pb filter=lfs diff=lfs merge=lfs -text
|
15 |
*.pt filter=lfs diff=lfs merge=lfs -text
|
16 |
*.pth filter=lfs diff=lfs merge=lfs -text
|
17 |
+
./minicoco.tar.gz filter=lfs diff=lfs merge=lfs -text
|
minicoco.py
ADDED
@@ -0,0 +1,531 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import os
|
2 |
+
import json
|
3 |
+
import logging
|
4 |
+
import datasets
|
5 |
+
import xml.etree.ElementTree as ET
|
6 |
+
from collections import defaultdict
|
7 |
+
|
8 |
+
|
9 |
+
_CITATION = """
|
10 |
+
MINICOCO2017
|
11 |
+
"""
|
12 |
+
|
13 |
+
_DESCRIPTION = """
|
14 |
+
MINICOCO2017
|
15 |
+
"""
|
16 |
+
|
17 |
+
_URLS = {
|
18 |
+
"minicoco2017": "minicoco.tar.gz"
|
19 |
+
}
|
20 |
+
|
21 |
+
# fmt: off
|
22 |
+
CLASS_INFOS = [
|
23 |
+
# name id train
|
24 |
+
('person', 1, 0),
|
25 |
+
('bicycle', 2, 1),
|
26 |
+
('car', 3, 2),
|
27 |
+
('motorcycle', 4, 3),
|
28 |
+
('airplane', 5, 4),
|
29 |
+
('bus', 6, 5),
|
30 |
+
('train', 7, 6),
|
31 |
+
('truck', 8, 7),
|
32 |
+
('boat', 9, 8),
|
33 |
+
('traffic light', 10, 9),
|
34 |
+
('fire hydrant', 11, 10),
|
35 |
+
('stop sign', 13, 11),
|
36 |
+
('parking meter', 14, 12),
|
37 |
+
('bench', 15, 13),
|
38 |
+
('bird', 16, 14),
|
39 |
+
('cat', 17, 15),
|
40 |
+
('dog', 18, 16),
|
41 |
+
('horse', 19, 17),
|
42 |
+
('sheep', 20, 18),
|
43 |
+
('cow', 21, 19),
|
44 |
+
('elephant', 22, 20),
|
45 |
+
('bear', 23, 21),
|
46 |
+
('zebra', 24, 22),
|
47 |
+
('giraffe', 25, 23),
|
48 |
+
('backpack', 27, 24),
|
49 |
+
('umbrella', 28, 25),
|
50 |
+
('handbag', 31, 26),
|
51 |
+
('tie', 32, 27),
|
52 |
+
('suitcase', 33, 28),
|
53 |
+
('frisbee', 34, 29),
|
54 |
+
('skis', 35, 30),
|
55 |
+
('snowboard', 36, 31),
|
56 |
+
('sports ball', 37, 32),
|
57 |
+
('kite', 38, 33),
|
58 |
+
('baseball bat', 39, 34),
|
59 |
+
('baseball glove', 40, 35),
|
60 |
+
('skateboard', 41, 36),
|
61 |
+
('surfboard', 42, 37),
|
62 |
+
('tennis racket', 43, 38),
|
63 |
+
('bottle', 44, 39),
|
64 |
+
('wine glass', 46, 40),
|
65 |
+
('cup', 47, 41),
|
66 |
+
('fork', 48, 42),
|
67 |
+
('knife', 49, 43),
|
68 |
+
('spoon', 50, 44),
|
69 |
+
('bowl', 51, 45),
|
70 |
+
('banana', 52, 46),
|
71 |
+
('apple', 53, 47),
|
72 |
+
('sandwich', 54, 48),
|
73 |
+
('orange', 55, 49),
|
74 |
+
('broccoli', 56, 50),
|
75 |
+
('carrot', 57, 51),
|
76 |
+
('hot dog', 58, 52),
|
77 |
+
('pizza', 59, 53),
|
78 |
+
('donut', 60, 54),
|
79 |
+
('cake', 61, 55),
|
80 |
+
('chair', 62, 56),
|
81 |
+
('couch', 63, 57),
|
82 |
+
('potted plant', 64, 58),
|
83 |
+
('bed', 65, 59),
|
84 |
+
('dining table', 67, 60),
|
85 |
+
('toilet', 70, 61),
|
86 |
+
('tv', 72, 62),
|
87 |
+
('laptop', 73, 63),
|
88 |
+
('mouse', 74, 64),
|
89 |
+
('remote', 75, 65),
|
90 |
+
('keyboard', 76, 66),
|
91 |
+
('cell phone', 77, 67),
|
92 |
+
('microwave', 78, 68),
|
93 |
+
('oven', 79, 69),
|
94 |
+
('toaster', 80, 70),
|
95 |
+
('sink', 81, 71),
|
96 |
+
('refrigerator', 82, 72),
|
97 |
+
('book', 84, 73),
|
98 |
+
('clock', 85, 74),
|
99 |
+
('vase', 86, 75),
|
100 |
+
('scissors', 87, 76),
|
101 |
+
('teddy bear', 88, 77),
|
102 |
+
('hair drier', 89, 78),
|
103 |
+
('toothbrush', 90, 79)
|
104 |
+
]
|
105 |
+
|
106 |
+
KEYPOINTS_INFOS=[
|
107 |
+
# name id train
|
108 |
+
# ('nose', 1, 0),
|
109 |
+
# ('left_eye', 2, 1),
|
110 |
+
# ('right_eye', 3, 2),
|
111 |
+
# ('left_ear', 4, 3),
|
112 |
+
# ('right_ear', 5, 4),
|
113 |
+
# ('left_shoulder', 6, 5),
|
114 |
+
# ('right_shoulder', 7, 6),
|
115 |
+
# ('left_elbow', 8, 7),
|
116 |
+
# ('right_elbow', 9, 8),
|
117 |
+
# ('left_wrist', 10, 9),
|
118 |
+
# ('right_wrist', 11, 10),
|
119 |
+
# ('left_hip', 12, 11),
|
120 |
+
# ('right_hip', 13, 12),
|
121 |
+
# ('left_knee', 14, 13),
|
122 |
+
# ('right_knee', 15, 14),
|
123 |
+
# ('left_ankle', 16, 15),
|
124 |
+
# ('right_ankle', 17, 16)
|
125 |
+
('none', 1, 0),
|
126 |
+
('nose', 2, 1),
|
127 |
+
('left_eye', 3, 2),
|
128 |
+
('right_eye', 4, 3),
|
129 |
+
('left_ear', 5, 4),
|
130 |
+
('right_ear', 6, 5),
|
131 |
+
('left_shoulder', 7, 6),
|
132 |
+
('right_shoulder', 8, 7),
|
133 |
+
('left_elbow', 9, 8),
|
134 |
+
('right_elbow', 10, 9),
|
135 |
+
('left_wrist', 11, 10),
|
136 |
+
('right_wrist', 12, 11),
|
137 |
+
('left_hip', 13, 12),
|
138 |
+
('right_hip', 14, 13),
|
139 |
+
('left_knee', 15, 14),
|
140 |
+
('right_knee', 16, 15),
|
141 |
+
('left_ankle', 17, 16),
|
142 |
+
('right_ankle', 18, 17)
|
143 |
+
]
|
144 |
+
|
145 |
+
|
146 |
+
# fmt: on
|
147 |
+
CLASS_NAMES = [CLASS_INFO[0] for CLASS_INFO in CLASS_INFOS]
|
148 |
+
KEYPOINTS_NAMES = [KEYPOINTS_INFO[0] for KEYPOINTS_INFO in KEYPOINTS_INFOS]
|
149 |
+
|
150 |
+
CLASS_DICT = {CLASS_INFO[0]: CLASS_INFO[2] for CLASS_INFO in CLASS_INFOS}
|
151 |
+
CATEGORY_ID2CLASS_NAMES = {CLASS_INFO[1]: CLASS_INFO[0] for CLASS_INFO in CLASS_INFOS}
|
152 |
+
KEYPOINTS_DICT = {KEYPOINTS_INFO[0]: KEYPOINTS_INFO[1] for KEYPOINTS_INFO in KEYPOINTS_INFOS}
|
153 |
+
|
154 |
+
|
155 |
+
# datasets.Features
|
156 |
+
detection_features = datasets.Features(
|
157 |
+
{
|
158 |
+
"id": datasets.Value("int32"),
|
159 |
+
"image": datasets.Value("string"),
|
160 |
+
"height": datasets.Value("int32"),
|
161 |
+
"width": datasets.Value("int32"),
|
162 |
+
"objects": datasets.features.Sequence(
|
163 |
+
{
|
164 |
+
"bboxes": datasets.Sequence(datasets.Value("float32")),
|
165 |
+
"classes": datasets.features.ClassLabel(names=CLASS_NAMES),
|
166 |
+
}
|
167 |
+
),
|
168 |
+
}
|
169 |
+
)
|
170 |
+
|
171 |
+
segmentation_features = datasets.Features(
|
172 |
+
{
|
173 |
+
"id": datasets.Value("int32"),
|
174 |
+
"image": datasets.Value("string"),
|
175 |
+
"height": datasets.Value("int32"),
|
176 |
+
"width": datasets.Value("int32"),
|
177 |
+
"objects": datasets.features.Sequence(
|
178 |
+
{
|
179 |
+
"bboxes": datasets.Sequence(datasets.Value("float32")),
|
180 |
+
"classes": datasets.features.ClassLabel(names=CLASS_NAMES),
|
181 |
+
'segmentation':datasets.Sequence(datasets.Value("float32")),
|
182 |
+
'iscrowd':datasets.Value("int32"),
|
183 |
+
}
|
184 |
+
),
|
185 |
+
}
|
186 |
+
)
|
187 |
+
|
188 |
+
captions_features = datasets.Features(
|
189 |
+
{
|
190 |
+
"id": datasets.Value("int32"),
|
191 |
+
"image": datasets.Value("string"),
|
192 |
+
"height": datasets.Value("int32"),
|
193 |
+
"width": datasets.Value("int32"),
|
194 |
+
"captions": datasets.features.Sequence(datasets.Value("string")),
|
195 |
+
}
|
196 |
+
)
|
197 |
+
|
198 |
+
keypoint_features = datasets.Features(
|
199 |
+
# 这里可能有点问题,因为模型的keypoint的标注的类别没别没有增加进来,
|
200 |
+
# 有点复杂,后面再finetune,现在基本信息已经正确
|
201 |
+
{
|
202 |
+
"id": datasets.Value("int32"),
|
203 |
+
"image": datasets.Value("string"),
|
204 |
+
"height": datasets.Value("int32"),
|
205 |
+
"width": datasets.Value("int32"),
|
206 |
+
"objects": datasets.features.Sequence(
|
207 |
+
{
|
208 |
+
"bboxes": datasets.Sequence(datasets.Value("float32")),
|
209 |
+
"classes": datasets.features.ClassLabel(names=CLASS_NAMES),
|
210 |
+
'keypoints':datasets.Sequence(datasets.Value("float32")),
|
211 |
+
"num_keypoints":datasets.Value("int32")
|
212 |
+
}
|
213 |
+
),
|
214 |
+
}
|
215 |
+
)
|
216 |
+
|
217 |
+
_DATASET_FEATURES = {
|
218 |
+
"detection": detection_features,
|
219 |
+
"segmentation":segmentation_features,
|
220 |
+
"caption": captions_features,
|
221 |
+
"keypoint": keypoint_features
|
222 |
+
}
|
223 |
+
|
224 |
+
|
225 |
+
def get_captions_annotation(captions_path):
|
226 |
+
with open(captions_path,'r') as f:
|
227 |
+
anno_captions = json.load(f)
|
228 |
+
|
229 |
+
anno_infos = defaultdict(list)
|
230 |
+
images_infos = list()
|
231 |
+
|
232 |
+
for caption_info in anno_captions['annotations']:
|
233 |
+
# caption_info={'image_id': 179765, 'id': 38, 'caption': 'A black Honda motorcycle parked in front of a garage.'}
|
234 |
+
caption = caption_info['caption']
|
235 |
+
image_id = caption_info['image_id']
|
236 |
+
|
237 |
+
anno_infos[image_id].append(caption)
|
238 |
+
|
239 |
+
for image in anno_captions['images']:
|
240 |
+
# image={'license': 4, 'file_name': '000000397133.jpg', 'coco_url': 'http://images.cocodataset.org/val2017/000000397133.jpg', 'height': 427, 'width': 640, 'date_captured': '2013-11-14 17:02:52', 'flickr_url': 'http://farm7.staticflickr.com/6116/6255196340_da26cf2c9e_z.jpg', 'id': 397133}
|
241 |
+
images_infos.append({
|
242 |
+
"image_name":image['file_name'],
|
243 |
+
"height": image["height"],
|
244 |
+
"width":image["width"],
|
245 |
+
"image_id":image['id']
|
246 |
+
})
|
247 |
+
|
248 |
+
return anno_infos, images_infos
|
249 |
+
|
250 |
+
|
251 |
+
def get_instances_annotation(instances_path):
|
252 |
+
with open(instances_path,'r') as f:
|
253 |
+
anno_instances = json.load(f)
|
254 |
+
|
255 |
+
anno_infos = dict()
|
256 |
+
images_infos = list()
|
257 |
+
|
258 |
+
for instance_info in anno_instances['annotations']:
|
259 |
+
# instance_info = {'segmentation': [[510.66, 423.01, 511.72, 420.03, 510.45, 416.0, 510.34, 413.02,
|
260 |
+
# 510.77, 410.26, 510.77, 407.5, 510.34, 405.16, 511.51, 402.83, 511.41, 400.49, 510.24, 398.16,
|
261 |
+
# 509.39, 397.31, 504.61, 399.22, 502.17, 399.64, 500.89, 401.66, 500.47, 402.08, 499.09, 401.87,
|
262 |
+
# 495.79, 401.98, 490.59, 401.77, 488.79, 401.77, 485.39, 398.58, 483.9, 397.31, 481.56, 396.35,
|
263 |
+
# 478.48, 395.93, 476.68, 396.03, 475.4, 396.77, 473.92, 398.79, 473.28, 399.96, 473.49, 401.87,
|
264 |
+
# 474.56, 403.47, 473.07, 405.59, 473.39, 407.71, 476.68, 409.41, 479.23, 409.73, 481.56, 410.69,
|
265 |
+
# 480.4, 411.85, 481.35, 414.93, 479.86, 418.65, 477.32, 420.03, 476.04, 422.58, 479.02, 422.58,
|
266 |
+
# 480.29, 423.01, 483.79, 419.93, 486.66, 416.21, 490.06, 415.57, 492.18, 416.85, 491.65, 420.24,
|
267 |
+
# 492.82, 422.9, 493.56, 424.39, 496.43, 424.6, 498.02, 423.01, 498.13, 421.31, 497.07, 420.03,
|
268 |
+
# 497.07, 415.15, 496.33, 414.51, 501.1, 411.96, 502.06, 411.32, 503.02, 415.04, 503.33, 418.12,
|
269 |
+
# 501.1, 420.24, 498.98, 421.63, 500.47, 424.39, 505.03, 423.32, 506.2, 421.31, 507.69, 419.5,
|
270 |
+
# 506.31, 423.32, 510.03, 423.01, 510.45, 423.01]], 'area': 702.1057499999998, 'iscrowd': 0,
|
271 |
+
# 'image_id': 289343, 'bbox': [473.07, 395.93, 38.65, 28.67], 'category_id': 18, 'id': 1768}
|
272 |
+
bbox = instance_info['bbox']
|
273 |
+
image_id = instance_info['image_id']
|
274 |
+
segmentation = instance_info['segmentation'][0]
|
275 |
+
|
276 |
+
if image_id in anno_infos:
|
277 |
+
anno_infos[image_id].append(
|
278 |
+
{
|
279 |
+
"segmentation": segmentation,
|
280 |
+
"bbox": bbox,
|
281 |
+
'iscrowd':instance_info['iscrowd'],
|
282 |
+
"classes":CATEGORY_ID2CLASS_NAMES[instance_info['category_id']]
|
283 |
+
}
|
284 |
+
)
|
285 |
+
else:
|
286 |
+
anno_infos[image_id]=[
|
287 |
+
{
|
288 |
+
"segmentation": segmentation,
|
289 |
+
"bbox": bbox,
|
290 |
+
'iscrowd':instance_info['iscrowd'],
|
291 |
+
"classes":CATEGORY_ID2CLASS_NAMES[instance_info['category_id']]
|
292 |
+
}
|
293 |
+
]
|
294 |
+
|
295 |
+
|
296 |
+
for image in anno_instances['images']:
|
297 |
+
# image={'license': 4, 'file_name': '000000397133.jpg',
|
298 |
+
# 'coco_url': 'http://images.cocodataset.org/val2017/000000397133.jpg',
|
299 |
+
# 'height': 427, 'width': 640, 'date_captured': '2013-11-14 17:02:52',
|
300 |
+
# 'flickr_url': 'http://farm7.staticflickr.com/6116/6255196340_da26cf2c9e_z.jpg', 'id': 397133}
|
301 |
+
images_infos.append({
|
302 |
+
"image_name":image['file_name'],
|
303 |
+
"height": image["height"],
|
304 |
+
"width":image["width"],
|
305 |
+
"image_id":image['id']
|
306 |
+
})
|
307 |
+
|
308 |
+
return anno_infos, images_infos
|
309 |
+
|
310 |
+
|
311 |
+
def get_keypoints_annotation(keypoints_path):
|
312 |
+
with open(keypoints_path,'r') as f:
|
313 |
+
anno_keypoints = json.load(f)
|
314 |
+
|
315 |
+
anno_infos = dict()
|
316 |
+
images_infos = list()
|
317 |
+
|
318 |
+
for keypoint_info in anno_keypoints['annotations']:
|
319 |
+
# keypoint_info = {'segmentation': [[63.2, 229.21, 65.73, 208.99, 70.79, 187.92, 78.37, 162.64, 84.27, 146.63, 84.27, 132.3, 75.84, 109.55, 90.17, 97.75, 104.49, 96.91, 114.61, 102.81, 123.88, 123.88, 137.36, 136.52, 153.37, 150.84, 146.63, 169.38, 144.1, 180.34, 142.42, 190.45, 137.36, 209.83, 139.89, 230.9, 128.09, 232.58, 97.75, 235.11, 81.74, 237.64, 87.64, 208.99, 85.96, 186.24, 78.37, 198.88, 75.84, 224.16, 68.26, 239.33, 60.67, 230.9]], 'num_keypoints': 12, 'area': 8096.3096, 'iscrowd': 0, 'keypoints': [100, 135, 2, 102, 127, 2, 94, 131, 2, 112, 121, 2, 91, 132, 2, 137, 148, 2, 81, 158, 2, 150, 179, 1, 76, 193, 2, 0, 0, 0, 70, 234, 2, 136, 242, 1, 104, 246, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0], 'image_id': 275749, 'bbox': [60.67, 96.91, 92.7, 142.42], 'category_id': 1, 'id': 232027}
|
320 |
+
bbox = keypoint_info['bbox']
|
321 |
+
image_id = keypoint_info['image_id']
|
322 |
+
|
323 |
+
if image_id in anno_infos:
|
324 |
+
anno_infos[image_id].append(
|
325 |
+
{
|
326 |
+
"bbox": bbox,
|
327 |
+
"classes":CATEGORY_ID2CLASS_NAMES[keypoint_info['category_id']],
|
328 |
+
'keypoints':keypoint_info['keypoints'],
|
329 |
+
"num_keypoints":keypoint_info['num_keypoints'],
|
330 |
+
}
|
331 |
+
)
|
332 |
+
else:
|
333 |
+
anno_infos[image_id]=[
|
334 |
+
{
|
335 |
+
"bbox": bbox,
|
336 |
+
"classes":CATEGORY_ID2CLASS_NAMES[keypoint_info['category_id']],
|
337 |
+
'keypoints':keypoint_info['keypoints'],
|
338 |
+
"num_keypoints":keypoint_info['num_keypoints'],
|
339 |
+
}
|
340 |
+
]
|
341 |
+
|
342 |
+
|
343 |
+
for image in anno_keypoints['images']:
|
344 |
+
# image={'license': 4, 'file_name': '000000397133.jpg', 'coco_url': 'http://images.cocodataset.org/val2017/000000397133.jpg', 'height': 427, 'width': 640, 'date_captured': '2013-11-14 17:02:52', 'flickr_url': 'http://farm7.staticflickr.com/6116/6255196340_da26cf2c9e_z.jpg', 'id': 397133}
|
345 |
+
images_infos.append({
|
346 |
+
"image_name":image['file_name'],
|
347 |
+
"height": image["height"],
|
348 |
+
"width":image["width"],
|
349 |
+
"image_id":image['id']
|
350 |
+
})
|
351 |
+
|
352 |
+
return anno_infos, images_infos
|
353 |
+
|
354 |
+
|
355 |
+
class MINICOCOConfig(datasets.BuilderConfig):
|
356 |
+
def __init__(self, data_name, task_name, **kwargs):
|
357 |
+
"""
|
358 |
+
|
359 |
+
Args:
|
360 |
+
**kwargs: keyword arguments forwarded to super.
|
361 |
+
"""
|
362 |
+
super().__init__(**kwargs)
|
363 |
+
assert data_name in ["minicoco2017"] and task_name in [
|
364 |
+
"detection",
|
365 |
+
"segmentation",
|
366 |
+
"caption",
|
367 |
+
"keypoint"
|
368 |
+
]
|
369 |
+
self.data_name = data_name
|
370 |
+
self.task_name = task_name
|
371 |
+
|
372 |
+
|
373 |
+
class PASCALDataset(datasets.GeneratorBasedBuilder):
|
374 |
+
|
375 |
+
BUILDER_CONFIGS = [
|
376 |
+
MINICOCOConfig(
|
377 |
+
name="minicoco2017_detection",
|
378 |
+
version=datasets.Version("1.0.0", ""),
|
379 |
+
description="minicoco2017 detection dataset",
|
380 |
+
data_name="minicoco2017",
|
381 |
+
task_name="detection",
|
382 |
+
),
|
383 |
+
MINICOCOConfig(
|
384 |
+
name="minicoco2017_segmentation",
|
385 |
+
version=datasets.Version("1.0.0", ""),
|
386 |
+
description="minicoco2017 segmentation dataset",
|
387 |
+
data_name="minicoco2017",
|
388 |
+
task_name="segmentation",
|
389 |
+
),
|
390 |
+
MINICOCOConfig(
|
391 |
+
name="minicoco2017_caption",
|
392 |
+
version=datasets.Version("1.0.0", ""),
|
393 |
+
description="minicoco2017 caption dataset",
|
394 |
+
data_name="minicoco2017",
|
395 |
+
task_name="caption",
|
396 |
+
),
|
397 |
+
MINICOCOConfig(
|
398 |
+
name="minicoco2017_keypoint",
|
399 |
+
version=datasets.Version("1.0.0", ""),
|
400 |
+
description="minicoco2017 keypoint dataset",
|
401 |
+
data_name="minicoco2017",
|
402 |
+
task_name="keypoint",
|
403 |
+
)
|
404 |
+
]
|
405 |
+
|
406 |
+
def _info(self):
|
407 |
+
return datasets.DatasetInfo(
|
408 |
+
description=_DESCRIPTION,
|
409 |
+
features=_DATASET_FEATURES[self.config.task_name],
|
410 |
+
# No default supervised_keys (as we have to pass both question
|
411 |
+
# and context as input).
|
412 |
+
supervised_keys=None,
|
413 |
+
homepage="https://fuliucansheng.github.io/",
|
414 |
+
citation=_CITATION,
|
415 |
+
)
|
416 |
+
|
417 |
+
def _split_generators(self, dl_manager):
|
418 |
+
downloaded_files = dl_manager.download_and_extract(_URLS[self.config.data_name])
|
419 |
+
|
420 |
+
return [
|
421 |
+
datasets.SplitGenerator(
|
422 |
+
name=datasets.Split.TRAIN,
|
423 |
+
gen_kwargs={"filepath": downloaded_files, "split": "train"},
|
424 |
+
),
|
425 |
+
datasets.SplitGenerator(
|
426 |
+
name=datasets.Split.VALIDATION,
|
427 |
+
gen_kwargs={"filepath": downloaded_files, "split": "val"},
|
428 |
+
),
|
429 |
+
datasets.SplitGenerator(
|
430 |
+
name=datasets.Split.TEST,
|
431 |
+
gen_kwargs={"filepath": downloaded_files, "split": "test"},
|
432 |
+
),
|
433 |
+
]
|
434 |
+
|
435 |
+
def _generate_examples(self, filepath, split):
|
436 |
+
"""This function returns the examples in the raw (text) form."""
|
437 |
+
|
438 |
+
# filepath = os.path.join(filepath, os.listdir(filepath)[0]) # mine add
|
439 |
+
|
440 |
+
logging.info("generating examples from = %s, split = %s", filepath, split)
|
441 |
+
task_name = self.config.task_name
|
442 |
+
|
443 |
+
if task_name == "caption":
|
444 |
+
captions_path = os.path.join(filepath, "annotations", "captions_" + split + "2017.json")
|
445 |
+
anno_infos, images_infos = get_captions_annotation(captions_path)
|
446 |
+
|
447 |
+
for id_, image in enumerate(images_infos):
|
448 |
+
image_path = os.path.join(filepath, split + "2017", image["image_name"])
|
449 |
+
if not os.path.exists(image_path):
|
450 |
+
continue
|
451 |
+
example = {
|
452 |
+
"id": id_,
|
453 |
+
"image": os.path.abspath(image_path),
|
454 |
+
"height": image["height"],
|
455 |
+
"width": image["width"],
|
456 |
+
"captions": anno_infos[image['image_id']],
|
457 |
+
}
|
458 |
+
yield id_, example
|
459 |
+
|
460 |
+
elif task_name=="detection":
|
461 |
+
instances_path = os.path.join(filepath, "annotations", "instances_" + split + "2017.json")
|
462 |
+
anno_infos, images_infos = get_instances_annotation(instances_path)
|
463 |
+
|
464 |
+
for id_, image in enumerate(images_infos):
|
465 |
+
image_path = os.path.join(filepath, split + "2017", image["image_name"])
|
466 |
+
if not os.path.exists(image_path):
|
467 |
+
continue
|
468 |
+
example = {
|
469 |
+
"id": id_,
|
470 |
+
"image": os.path.abspath(image_path),
|
471 |
+
"height": image["height"],
|
472 |
+
"width": image["width"],
|
473 |
+
"objects":[
|
474 |
+
{
|
475 |
+
"bboxes": object_info["bbox"],
|
476 |
+
"classes": object_info["classes"]
|
477 |
+
}
|
478 |
+
for object_info in anno_infos[image['image_id']]
|
479 |
+
]
|
480 |
+
}
|
481 |
+
yield id_, example
|
482 |
+
|
483 |
+
elif task_name=="segmentation":
|
484 |
+
instances_path = os.path.join(filepath, "annotations", "instances_" + split + "2017.json")
|
485 |
+
anno_infos, images_infos = get_instances_annotation(instances_path)
|
486 |
+
|
487 |
+
for id_, image in enumerate(images_infos):
|
488 |
+
image_path = os.path.join(filepath, split + "2017", image["image_name"])
|
489 |
+
if not os.path.exists(image_path):
|
490 |
+
continue
|
491 |
+
example = {
|
492 |
+
"id": id_,
|
493 |
+
"image": os.path.abspath(image_path),
|
494 |
+
"height": image["height"],
|
495 |
+
"width": image["width"],
|
496 |
+
"objects":[
|
497 |
+
{
|
498 |
+
"bboxes": object_info["bbox"],
|
499 |
+
"classes": object_info["classes"],
|
500 |
+
'segmentation':object_info['segmentation'],
|
501 |
+
'iscrowd':object_info['iscrowd']
|
502 |
+
}
|
503 |
+
for object_info in anno_infos[image['image_id']]
|
504 |
+
]
|
505 |
+
}
|
506 |
+
yield id_, example
|
507 |
+
|
508 |
+
elif task_name=="keypoint":
|
509 |
+
keypoints_path = os.path.join(filepath, "annotations", "person_keypoints_" + split + "2017.json")
|
510 |
+
anno_infos, images_infos = get_keypoints_annotation(keypoints_path)
|
511 |
+
|
512 |
+
for id_, image in enumerate(images_infos):
|
513 |
+
image_path = os.path.join(filepath, split + "2017", image["image_name"])
|
514 |
+
if not os.path.exists(image_path):
|
515 |
+
continue
|
516 |
+
example = {
|
517 |
+
"id": id_,
|
518 |
+
"image": os.path.abspath(image_path),
|
519 |
+
"height": image["height"],
|
520 |
+
"width": image["width"],
|
521 |
+
"objects":[
|
522 |
+
{
|
523 |
+
"bboxes": object_info["bbox"],
|
524 |
+
"classes": object_info["classes"],
|
525 |
+
'keypoints':object_info['keypoints'],
|
526 |
+
"num_keypoints":object_info["num_keypoints"]
|
527 |
+
}
|
528 |
+
for object_info in anno_infos[image['image_id']]
|
529 |
+
]
|
530 |
+
}
|
531 |
+
yield id_, example
|
minicoco.tar.gz
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:347d6613f2564d33029f78fb52eebdaf4036013274b4cd9daa2aa52fb0e59fcf
|
3 |
+
size 6792618279
|