-
Notifications
You must be signed in to change notification settings - Fork 0
/
run_prior.py
289 lines (272 loc) · 10.6 KB
/
run_prior.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
import argparse
import math
import os
from pprint import pprint
from refer360_utils import test_get_nears
from refer360_utils import generate_baseline_cooccurrences
from refer360_utils import get_visualgenome_stats
from refer360_utils import get_spatialsense_stats
from refer360_utils import get_wordnet_stats
from refer360_utils import get_dataset_stats
from refer360_utils import evaluate_fov_caches
from refer360_utils import dump_fov_caches
from refer360_utils import dump_oracle_caches
from refer360_utils import dump_fov_stats
from refer360_utils import dump_gaussian_caches
parser = argparse.ArgumentParser()
parser.add_argument("--data_prefix",
choices=['td', 'refer360', 'r360tiny'],
default='refer360',
help='dataset prefix, default: refer360')
parser.add_argument('--angle_inc', type=int,
choices=[30, 45, 60],
default=30,
help='degrees between fovs, default=30')
parser.add_argument('--output_root', type=str,
default='./img_features',
help='Dump folder path like default=./img_features')
parser.add_argument('--diag_mode', type=int,
choices=[-1, 0, 1],
default=-1,
help='diagonal of cooccurrence data, default=-1')
parser.add_argument('--img_features_root', type=str,
default='./img_features',
help='img_features_root, default="./img_features"')
parser.add_argument('--obj_dict_file', type=str,
default='./tasks/FAST/data/vg_object_dictionaries.all.json',
help='obj_dict_file, default=./tasks/FAST/data/vg_object_dictionaries.all.json')
parser.add_argument('--word_embedding_path', type=str,
default='./tasks/FAST/data/cc.en.300.vec',
help='word embedding path, default=./tasks/FAST/data/cc.en.300.vec')
parser.add_argument('--cooccurrence_path', type=str,
default='./cooccurrences',
help='cooccurrence files root, default="./cooccurrences"')
parser.add_argument('--stats_path', type=str,
default='./stats',
help='stats files root, default="./stats"')
parser.add_argument('--version', type=str,
default='v5',
help='version of prior methods, default="v5"')
parser.add_argument('--prefix_iou', type=int, default=10,
help='prefix_iou, default=10')
parser.add_argument("--cooccurrence_method",
choices=['all', 'lm', 'language',
'vision', 'word_emb', 'dataset',
'baseline', 'custom'],
default='all',
help='prior method: default=all')
parser.add_argument('--test_get_nears', action='store_true')
parser.add_argument('--baseline_cooccurrences', action='store_true')
parser.add_argument('--visualgenome', action='store_true')
parser.add_argument('--spatialsense', action='store_true')
parser.add_argument('--wordnet', action='store_true')
parser.add_argument('--dataset_stats', action='store_true')
parser.add_argument('--evaluate_fov_caches', action='store_true')
parser.add_argument('--dump_fov_caches', action='store_true')
parser.add_argument('--dump_oracle_caches', action='store_true')
parser.add_argument('--dump_fov_stats', action='store_true')
parser.add_argument('--dump_gaussian_caches', action='store_true')
CAT2METHOD = {
'all': ['wn',
'word2vec',
'glove',
'fastText',
'gpt',
'gpt2',
'gpt3',
'gptneo',
'roberta',
'xlm',
'ss',
'vg'],
'language': ['wn',
'word2vec',
'glove',
'fastText',
'gpt',
'gpt2',
'gpt3',
'gptneo',
'roberta',
'xlm'],
'vision': ['ss',
'vg'],
'lm': ['gpt',
'gpt2',
'gpt3',
'gptneo',
'roberta',
'xlm'],
'word_emb': ['word2vec',
'glove',
'fastText']
}
def get_cooccurrence_files(method, cooccurrence_path, data_stats, version):
cooccurrence_files = []
if method == 'custom':
custom = [
'wn_v5',
'vg_v5',
'ss_v5']
for c in custom:
cooccurrence_files += [os.path.join(
cooccurrence_path, 'cooccurrence.{}.npy'.format(c))]
return cooccurrence_files
if method in CAT2METHOD:
for m in CAT2METHOD[method]:
cooccurrence_files += [os.path.join(
cooccurrence_path, 'cooccurrence.{}_{}.npy'.format(m, version))]
if method == 'all' or method == 'dataset':
data_stats_method = os.path.join(
cooccurrence_path, 'cooccurrence.{}_{}.npy'.format(data_stats, version))
cooccurrence_files += [data_stats_method]
if method == 'all' or method == 'baseline':
baselines = ['uniform',
'random100',
'diagonal'
]
for baseline in baselines:
cooccurrence_files += [os.path.join(
cooccurrence_path, 'cooccurrence.{}.npy'.format(baseline))]
return cooccurrence_files
args = parser.parse_args()
data_prefix = args.data_prefix
angle_inc = args.angle_inc
output_root = args.output_root
diag_mode = args.diag_mode
version = args.version
prefix_iou = args.prefix_iou
obj_dict_file = args.obj_dict_file
word_embedding_path = args.word_embedding_path
cooccurrence_path = args.cooccurrence_path
stats_path = args.stats_path
n_fovs = int((360 / angle_inc)*math.ceil(150/angle_inc))
butd_filename = './img_features/{}_{}degrees_obj36.tsv'.format(
data_prefix,
angle_inc)
cache_root = '{}_data/cached_data_{}degrees'.format(data_prefix, angle_inc)
image_list_file = './{}_data/imagelist.txt'.format(data_prefix)
iou = float('0.{}'.format(prefix_iou))
msuffix = 'mIOU{}'.format(prefix_iou) # 'mHALF'
print('angle_inc', angle_inc)
print('n_fovs', n_fovs)
print('butd_filename', butd_filename)
print('cache_root', cache_root)
print('image_list_file', image_list_file)
print('output_root', output_root)
print('version:', version)
print('msuffix:', msuffix)
print('iou:', iou)
print('diag_mode:', diag_mode)
if args.test_get_nears:
test_get_nears()
if args.baseline_cooccurrences:
print('running generate_baseline_cooccurrences()')
generate_baseline_cooccurrences(obj_dict_file=obj_dict_file,
cooccurrence_path=cooccurrence_path)
if args.visualgenome:
print('running get_visualgenome_stats()')
get_visualgenome_stats(obj_dict_file=obj_dict_file,
cooccurrence_path=cooccurrence_path,
version=version)
if args.spatialsense:
print('running get_spatialsense_stats()')
get_spatialsense_stats(obj_dict_file=obj_dict_file,
cooccurrence_path=cooccurrence_path,
version=version)
if args.wordnet:
print('running get_wordnet_stats()')
get_wordnet_stats(obj_dict_file=obj_dict_file,
cooccurrence_path=cooccurrence_path,
version=version)
if args.dataset_stats:
print('running get_dataset_stats()')
get_dataset_stats(butd_filename=butd_filename,
image_list_file=image_list_file,
n_fovs=n_fovs,
angle_inc=angle_inc,
data_prefix=data_prefix,
obj_dict_file=obj_dict_file,
cooccurrence_path=cooccurrence_path,
version=version)
data_stats = '{}_d{}_butd'.format(data_prefix, angle_inc)
cooccurrence_files = get_cooccurrence_files(
args.cooccurrence_method, cooccurrence_path, data_stats, version)
print('cooccurrence_files:')
pprint(cooccurrence_files)
if args.evaluate_fov_caches:
print('running evaluate_fov_caches()')
evaluate_fov_caches(
cache_root=cache_root,
butd_filename=butd_filename,
image_list_file=image_list_file,
n_fovs=n_fovs,
angle_inc=angle_inc,
data_prefix=data_prefix,
word_embedding_path=word_embedding_path,
obj_dict_file=obj_dict_file,
cooccurrence_files=cooccurrence_files,
output_root=output_root,
diag_mode=diag_mode,
msuffix=msuffix)
if args.dump_fov_caches:
print('running dump_fov_caches()')
dump_fov_caches(butd_filename=butd_filename,
image_list_file=image_list_file,
n_fovs=n_fovs,
angle_inc=angle_inc,
data_prefix=data_prefix,
word_embedding_path=word_embedding_path,
obj_dict_file=obj_dict_file,
cooccurrence_files=cooccurrence_files,
output_root=output_root,
diag_mode=diag_mode,
msuffix=msuffix)
if args.dump_oracle_caches:
print('running dump_oracle_caches()')
dump_oracle_caches(
cache_root=cache_root,
butd_filename=butd_filename,
image_list_file=image_list_file,
n_fovs=n_fovs,
angle_inc=angle_inc,
data_prefix=data_prefix,
word_embedding_path=word_embedding_path,
obj_dict_file=obj_dict_file,
output_root=output_root)
if args.dump_fov_stats:
print('running dump_fov_stats()')
stats_files = [
'./stats/{}_cached{}degrees_stats.npy'.format(data_prefix, angle_inc)
]
outfiles = [
'img_features/{}_{}degrees_r{}statsall_v3.tsv'.format(
data_prefix, angle_inc, angle_inc),
]
methods = [
'all_regions',
]
dump_fov_stats(butd_filename=butd_filename,
image_list_file=image_list_file,
n_fovs=n_fovs,
angle_inc=angle_inc,
word_embedding_path=word_embedding_path,
obj_dict_file=obj_dict_file,
stats_files=stats_files,
outfiles=outfiles,
methods=methods)
if args.dump_gaussian_caches:
data_stats_method = os.path.join(
cooccurrence_path, 'cooccurrence.{}_{}.npy'.format(data_stats, version))
cooccurrence_files = [data_stats_method]
msuffix = 'GAUSSIAN'
dump_gaussian_caches(butd_filename=butd_filename,
image_list_file=image_list_file,
n_fovs=n_fovs,
angle_inc=angle_inc,
data_prefix=data_prefix,
word_embedding_path=word_embedding_path,
obj_dict_file=obj_dict_file,
output_root=output_root,
cooccurrence_files=cooccurrence_files,
msuffix=msuffix)