kimsan0622
commited on
Commit
•
6b0a878
1
Parent(s):
2015ff6
Update coco.py
Browse files
coco.py
CHANGED
@@ -157,6 +157,118 @@ CAT = [
|
|
157 |
"hair brush",
|
158 |
]
|
159 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
160 |
SUPER_CAT = [
|
161 |
"none",
|
162 |
"person",
|
@@ -173,6 +285,24 @@ SUPER_CAT = [
|
|
173 |
"indoor",
|
174 |
]
|
175 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
176 |
CAT2SUPER_CAT = [
|
177 |
"none",
|
178 |
"person",
|
@@ -266,13 +396,118 @@ CAT2SUPER_CAT = [
|
|
266 |
"indoor",
|
267 |
"indoor",
|
268 |
"indoor",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
269 |
]
|
270 |
|
271 |
|
272 |
-
|
273 |
-
|
274 |
-
|
275 |
-
|
276 |
class AnnotationType(object):
|
277 |
"""Enum of the annotation format types.
|
278 |
Splits are annotated with different formats.
|
@@ -313,8 +548,8 @@ PANOPTIC_FEATURE = datasets.Features(
|
|
313 |
"bbox": datasets.Sequence(
|
314 |
feature=datasets.Value("float32")
|
315 |
),
|
316 |
-
"label": datasets.ClassLabel(names=
|
317 |
-
"super_cat_label": datasets.ClassLabel(names=
|
318 |
"is_crowd": datasets.Value("bool"),
|
319 |
})),
|
320 |
"panoptic_image": datasets.Image(),
|
@@ -605,10 +840,10 @@ class Coco(datasets.GeneratorBasedBuilder):
|
|
605 |
# pylint: disable=cell-var-from-loop
|
606 |
# build_bbox is only used within the loop so it is ok to use image_info
|
607 |
return [
|
608 |
-
y,
|
609 |
x,
|
610 |
-
|
611 |
(x + width),
|
|
|
612 |
]
|
613 |
# pylint: enable=cell-var-from-loop
|
614 |
|
@@ -622,7 +857,7 @@ class Coco(datasets.GeneratorBasedBuilder):
|
|
622 |
'area': instance['area'],
|
623 |
'bbox': build_bbox(*instance['bbox']),
|
624 |
'label': instance['category_id'],
|
625 |
-
'super_cat_label':
|
626 |
'is_crowd': bool(instance['iscrowd']),
|
627 |
}
|
628 |
for instance in instances
|
|
|
157 |
"hair brush",
|
158 |
]
|
159 |
|
160 |
+
CAT_PANOPTIC = CAT + [
|
161 |
+
"banner",
|
162 |
+
"blanket",
|
163 |
+
"none1",
|
164 |
+
"bridge",
|
165 |
+
"none2",
|
166 |
+
"none3",
|
167 |
+
"none4",
|
168 |
+
"none5",
|
169 |
+
"cardboard",
|
170 |
+
"none6",
|
171 |
+
"none7",
|
172 |
+
"none8",
|
173 |
+
"none9",
|
174 |
+
"none10",
|
175 |
+
"none11",
|
176 |
+
"counter",
|
177 |
+
"none12",
|
178 |
+
"curtain",
|
179 |
+
"none13",
|
180 |
+
"none14",
|
181 |
+
"door-stuff",
|
182 |
+
"none15",
|
183 |
+
"none16",
|
184 |
+
"none17",
|
185 |
+
"none18",
|
186 |
+
"none19",
|
187 |
+
"floor-wood",
|
188 |
+
"flower",
|
189 |
+
"none20",
|
190 |
+
"none21",
|
191 |
+
"fruit",
|
192 |
+
"none22",
|
193 |
+
"none23",
|
194 |
+
"gravel",
|
195 |
+
"none24",
|
196 |
+
"none25",
|
197 |
+
"house",
|
198 |
+
"none26",
|
199 |
+
"light",
|
200 |
+
"none27",
|
201 |
+
"none28",
|
202 |
+
"mirror-stuff",
|
203 |
+
"none29",
|
204 |
+
"none30",
|
205 |
+
"none31",
|
206 |
+
"none32",
|
207 |
+
"net",
|
208 |
+
"none33",
|
209 |
+
"none34",
|
210 |
+
"pillow",
|
211 |
+
"none35",
|
212 |
+
"none36",
|
213 |
+
"platform",
|
214 |
+
"playingfield",
|
215 |
+
"none37",
|
216 |
+
"railroad",
|
217 |
+
"river",
|
218 |
+
"road",
|
219 |
+
"none38",
|
220 |
+
"roof",
|
221 |
+
"none39",
|
222 |
+
"none40",
|
223 |
+
"sand",
|
224 |
+
"sea",
|
225 |
+
"shelf",
|
226 |
+
"none41",
|
227 |
+
"none42",
|
228 |
+
"snow",
|
229 |
+
"none43",
|
230 |
+
"stairs",
|
231 |
+
"none44",
|
232 |
+
"none45",
|
233 |
+
"none46",
|
234 |
+
"none47",
|
235 |
+
"tent",
|
236 |
+
"none48",
|
237 |
+
"towel",
|
238 |
+
"none49",
|
239 |
+
"none50",
|
240 |
+
"wall-brick",
|
241 |
+
"none51",
|
242 |
+
"none52",
|
243 |
+
"none53",
|
244 |
+
"wall-stone",
|
245 |
+
"wall-tile",
|
246 |
+
"wall-wood",
|
247 |
+
"water-other",
|
248 |
+
"none54",
|
249 |
+
"window-blind",
|
250 |
+
"window-other",
|
251 |
+
"none55",
|
252 |
+
"none56",
|
253 |
+
"tree-merged",
|
254 |
+
"fence-merged",
|
255 |
+
"ceiling-merged",
|
256 |
+
"sky-other-merged",
|
257 |
+
"cabinet-merged",
|
258 |
+
"table-merged",
|
259 |
+
"floor-other-merged",
|
260 |
+
"pavement-merged",
|
261 |
+
"mountain-merged",
|
262 |
+
"grass-merged",
|
263 |
+
"dirt-merged",
|
264 |
+
"paper-merged",
|
265 |
+
"food-other-merged",
|
266 |
+
"building-other-merged",
|
267 |
+
"rock-merged",
|
268 |
+
"wall-other-merged",
|
269 |
+
"rug-merged",
|
270 |
+
]
|
271 |
+
|
272 |
SUPER_CAT = [
|
273 |
"none",
|
274 |
"person",
|
|
|
285 |
"indoor",
|
286 |
]
|
287 |
|
288 |
+
SUPER_CAT_PANOPTIC = SUPER_CAT + [
|
289 |
+
"textile",
|
290 |
+
"building",
|
291 |
+
"raw-material",
|
292 |
+
"furniture-stuff",
|
293 |
+
"floor",
|
294 |
+
"plant",
|
295 |
+
"food-stuff",
|
296 |
+
"ground",
|
297 |
+
"structural",
|
298 |
+
"water",
|
299 |
+
"wall",
|
300 |
+
"window",
|
301 |
+
"ceiling",
|
302 |
+
"sky",
|
303 |
+
"solid",
|
304 |
+
]
|
305 |
+
|
306 |
CAT2SUPER_CAT = [
|
307 |
"none",
|
308 |
"person",
|
|
|
396 |
"indoor",
|
397 |
"indoor",
|
398 |
"indoor",
|
399 |
+
"textile",
|
400 |
+
"textile",
|
401 |
+
"none",
|
402 |
+
"building",
|
403 |
+
"none",
|
404 |
+
"none",
|
405 |
+
"none",
|
406 |
+
"none",
|
407 |
+
"raw-material",
|
408 |
+
"none",
|
409 |
+
"none",
|
410 |
+
"none",
|
411 |
+
"none",
|
412 |
+
"none",
|
413 |
+
"none",
|
414 |
+
"furniture-stuff",
|
415 |
+
"none",
|
416 |
+
"textile",
|
417 |
+
"none",
|
418 |
+
"none",
|
419 |
+
"furniture-stuff",
|
420 |
+
"none",
|
421 |
+
"none",
|
422 |
+
"none",
|
423 |
+
"none",
|
424 |
+
"none",
|
425 |
+
"floor",
|
426 |
+
"plant",
|
427 |
+
"none",
|
428 |
+
"none",
|
429 |
+
"food-stuff",
|
430 |
+
"none",
|
431 |
+
"none",
|
432 |
+
"ground",
|
433 |
+
"none",
|
434 |
+
"none",
|
435 |
+
"building",
|
436 |
+
"none",
|
437 |
+
"furniture-stuff",
|
438 |
+
"none",
|
439 |
+
"none",
|
440 |
+
"furniture-stuff",
|
441 |
+
"none",
|
442 |
+
"none",
|
443 |
+
"none",
|
444 |
+
"none",
|
445 |
+
"structural",
|
446 |
+
"none",
|
447 |
+
"none",
|
448 |
+
"textile",
|
449 |
+
"none",
|
450 |
+
"none",
|
451 |
+
"ground",
|
452 |
+
"ground",
|
453 |
+
"none",
|
454 |
+
"ground",
|
455 |
+
"water",
|
456 |
+
"ground",
|
457 |
+
"none",
|
458 |
+
"building",
|
459 |
+
"none",
|
460 |
+
"none",
|
461 |
+
"ground",
|
462 |
+
"water",
|
463 |
+
"furniture-stuff",
|
464 |
+
"none",
|
465 |
+
"none",
|
466 |
+
"ground",
|
467 |
+
"none",
|
468 |
+
"furniture-stuff",
|
469 |
+
"none",
|
470 |
+
"none",
|
471 |
+
"none",
|
472 |
+
"none",
|
473 |
+
"building",
|
474 |
+
"none",
|
475 |
+
"textile",
|
476 |
+
"none",
|
477 |
+
"none",
|
478 |
+
"wall",
|
479 |
+
"none",
|
480 |
+
"none",
|
481 |
+
"none",
|
482 |
+
"wall",
|
483 |
+
"wall",
|
484 |
+
"wall",
|
485 |
+
"water",
|
486 |
+
"none",
|
487 |
+
"window",
|
488 |
+
"window",
|
489 |
+
"none",
|
490 |
+
"none",
|
491 |
+
"plant",
|
492 |
+
"structural",
|
493 |
+
"ceiling",
|
494 |
+
"sky",
|
495 |
+
"furniture-stuff",
|
496 |
+
"furniture-stuff",
|
497 |
+
"floor",
|
498 |
+
"ground",
|
499 |
+
"solid",
|
500 |
+
"plant",
|
501 |
+
"ground",
|
502 |
+
"raw-material",
|
503 |
+
"food-stuff",
|
504 |
+
"building",
|
505 |
+
"solid",
|
506 |
+
"wall",
|
507 |
+
"textile",
|
508 |
]
|
509 |
|
510 |
|
|
|
|
|
|
|
|
|
511 |
class AnnotationType(object):
|
512 |
"""Enum of the annotation format types.
|
513 |
Splits are annotated with different formats.
|
|
|
548 |
"bbox": datasets.Sequence(
|
549 |
feature=datasets.Value("float32")
|
550 |
),
|
551 |
+
"label": datasets.ClassLabel(names=CAT_PANOPTIC),
|
552 |
+
"super_cat_label": datasets.ClassLabel(names=SUPER_CAT_PANOPTIC),
|
553 |
"is_crowd": datasets.Value("bool"),
|
554 |
})),
|
555 |
"panoptic_image": datasets.Image(),
|
|
|
840 |
# pylint: disable=cell-var-from-loop
|
841 |
# build_bbox is only used within the loop so it is ok to use image_info
|
842 |
return [
|
|
|
843 |
x,
|
844 |
+
y,
|
845 |
(x + width),
|
846 |
+
(y + height),
|
847 |
]
|
848 |
# pylint: enable=cell-var-from-loop
|
849 |
|
|
|
857 |
'area': instance['area'],
|
858 |
'bbox': build_bbox(*instance['bbox']),
|
859 |
'label': instance['category_id'],
|
860 |
+
'super_cat_label': SUPER_CAT_PANOPTIC.index(CAT2SUPER_CAT[instance['category_id']]),
|
861 |
'is_crowd': bool(instance['iscrowd']),
|
862 |
}
|
863 |
for instance in instances
|