sushruthb commited on
Commit
c603d61
·
verified ·
1 Parent(s): 4549518

Upload folder using huggingface_hub

Browse files
._SweMPer-layout-lite.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:be37ea73332d2b169b5df7e9e94189963204774bb36962c8315cdace36a6a368
3
+ size 4096
._config_mask_rcnn_resized.yaml ADDED
Binary file (4.1 kB). View file
 
SweMPer-layout-lite.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a14db140867d80933082e7620e3c18576b33bc4c9187981bd6a1530a756014c8
3
+ size 846961818
config_mask_rcnn_resized.yaml ADDED
@@ -0,0 +1,320 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ CUDNN_BENCHMARK: false
2
+ DATALOADER:
3
+ ASPECT_RATIO_GROUPING: true
4
+ FILTER_EMPTY_ANNOTATIONS: true
5
+ NUM_WORKERS: 4
6
+ REPEAT_THRESHOLD: 0.0
7
+ SAMPLER_TRAIN: TrainingSampler
8
+ DATASETS:
9
+ PRECOMPUTED_PROPOSAL_TOPK_TEST: 1000
10
+ PRECOMPUTED_PROPOSAL_TOPK_TRAIN: 2000
11
+ PROPOSAL_FILES_TEST: []
12
+ PROPOSAL_FILES_TRAIN: []
13
+ TEST:
14
+ - publaynet-val
15
+ TRAIN:
16
+ - publaynet-train
17
+ GLOBAL:
18
+ HACK: 1.0
19
+ INPUT:
20
+ CROP:
21
+ ENABLED: false
22
+ SIZE:
23
+ - 0.9
24
+ - 0.9
25
+ TYPE: relative_range
26
+ FORMAT: BGR
27
+ MASK_FORMAT: polygon
28
+ MAX_SIZE_TEST: 1333
29
+ MAX_SIZE_TRAIN: 1333
30
+ MIN_SIZE_TEST: 800
31
+ MIN_SIZE_TRAIN:
32
+ - 50
33
+ - 100
34
+ - 200
35
+ - 300
36
+ - 400
37
+ - 500
38
+ - 640
39
+ - 672
40
+ - 704
41
+ - 736
42
+ - 768
43
+ - 800
44
+ - 1000
45
+ MIN_SIZE_TRAIN_SAMPLING: choice
46
+ MODEL:
47
+ ANCHOR_GENERATOR:
48
+ ANGLES:
49
+ - - -90
50
+ - 0
51
+ - 90
52
+ ASPECT_RATIOS:
53
+ - - 0.5
54
+ - 1.0
55
+ - 2.0
56
+ NAME: DefaultAnchorGenerator
57
+ OFFSET: 0.0
58
+ SIZES:
59
+ - - 4
60
+ - - 16
61
+ - - 64
62
+ - - 256
63
+ - - 1024
64
+ BACKBONE:
65
+ FREEZE_AT: 3
66
+ NAME: build_resnet_fpn_backbone
67
+ DEVICE: cuda
68
+ FPN:
69
+ FUSE_TYPE: sum
70
+ IN_FEATURES:
71
+ - res2
72
+ - res3
73
+ - res4
74
+ - res5
75
+ NORM: ''
76
+ OUT_CHANNELS: 256
77
+ KEYPOINT_ON: false
78
+ LOAD_PROPOSALS: false
79
+ MASK_ON: true
80
+ META_ARCHITECTURE: GeneralizedRCNN
81
+ PANOPTIC_FPN:
82
+ COMBINE:
83
+ ENABLED: true
84
+ INSTANCES_CONFIDENCE_THRESH: 0.5
85
+ OVERLAP_THRESH: 0.5
86
+ STUFF_AREA_LIMIT: 4096
87
+ INSTANCE_LOSS_WEIGHT: 1.0
88
+ PIXEL_MEAN:
89
+ - 103.53
90
+ - 116.28
91
+ - 123.675
92
+ PIXEL_STD:
93
+ - 57.375
94
+ - 57.12
95
+ - 58.395
96
+ PROPOSAL_GENERATOR:
97
+ MIN_SIZE: 0
98
+ NAME: RPN
99
+ RESNETS:
100
+ DEFORM_MODULATED: false
101
+ DEFORM_NUM_GROUPS: 1
102
+ DEFORM_ON_PER_STAGE:
103
+ - false
104
+ - false
105
+ - false
106
+ - false
107
+ DEPTH: 101
108
+ NORM: FrozenBN
109
+ NUM_GROUPS: 32
110
+ OUT_FEATURES:
111
+ - res2
112
+ - res3
113
+ - res4
114
+ - res5
115
+ RES2_OUT_CHANNELS: 256
116
+ RES5_DILATION: 1
117
+ STEM_OUT_CHANNELS: 64
118
+ STRIDE_IN_1X1: false
119
+ WIDTH_PER_GROUP: 8
120
+ RETINANET:
121
+ BBOX_REG_WEIGHTS:
122
+ - 1.0
123
+ - 1.0
124
+ - 1.0
125
+ - 1.0
126
+ FOCAL_LOSS_ALPHA: 0.25
127
+ FOCAL_LOSS_GAMMA: 2.0
128
+ IN_FEATURES:
129
+ - p3
130
+ - p4
131
+ - p5
132
+ - p6
133
+ - p7
134
+ IOU_LABELS:
135
+ - 0
136
+ - -1
137
+ - 1
138
+ IOU_THRESHOLDS:
139
+ - 0.4
140
+ - 0.5
141
+ NMS_THRESH_TEST: 0.5
142
+ NUM_CLASSES: 80
143
+ NUM_CONVS: 4
144
+ PRIOR_PROB: 0.01
145
+ SCORE_THRESH_TEST: 0.05
146
+ SMOOTH_L1_LOSS_BETA: 0.1
147
+ TOPK_CANDIDATES_TEST: 1000
148
+ ROI_BOX_CASCADE_HEAD:
149
+ BBOX_REG_WEIGHTS:
150
+ - - 10.0
151
+ - 10.0
152
+ - 5.0
153
+ - 5.0
154
+ - - 20.0
155
+ - 20.0
156
+ - 10.0
157
+ - 10.0
158
+ - - 30.0
159
+ - 30.0
160
+ - 15.0
161
+ - 15.0
162
+ IOUS:
163
+ - 0.5
164
+ - 0.6
165
+ - 0.7
166
+ ROI_BOX_HEAD:
167
+ BBOX_REG_WEIGHTS:
168
+ - 10.0
169
+ - 10.0
170
+ - 5.0
171
+ - 5.0
172
+ CLS_AGNOSTIC_BBOX_REG: false
173
+ CONV_DIM: 256
174
+ FC_DIM: 1024
175
+ NAME: FastRCNNConvFCHead
176
+ NORM: ''
177
+ NUM_CONV: 0
178
+ NUM_FC: 2
179
+ POOLER_RESOLUTION: 7
180
+ POOLER_SAMPLING_RATIO: 0
181
+ POOLER_TYPE: ROIAlignV2
182
+ SMOOTH_L1_BETA: 0.0
183
+ TRAIN_ON_PRED_BOXES: false
184
+ ROI_HEADS:
185
+ BATCH_SIZE_PER_IMAGE: 512
186
+ IN_FEATURES:
187
+ - p2
188
+ - p3
189
+ - p4
190
+ - p5
191
+ IOU_LABELS:
192
+ - 0
193
+ - 1
194
+ IOU_THRESHOLDS:
195
+ - 0.5
196
+ NAME: StandardROIHeads
197
+ NMS_THRESH_TEST: 0.5
198
+ NUM_CLASSES: 9
199
+ POSITIVE_FRACTION: 0.25
200
+ PROPOSAL_APPEND_GT: true
201
+ SCORE_THRESH_TEST: 0.05
202
+ ROI_KEYPOINT_HEAD:
203
+ CONV_DIMS:
204
+ - 1200
205
+ - 1200
206
+ - 1200
207
+ - 1200
208
+ - 1200
209
+ - 1200
210
+ - 1200
211
+ - 1200
212
+ LOSS_WEIGHT: 1.0
213
+ MIN_KEYPOINTS_PER_IMAGE: 1
214
+ NAME: KRCNNConvDeconvUpsampleHead
215
+ NORMALIZE_LOSS_BY_VISIBLE_KEYPOINTS: true
216
+ NUM_KEYPOINTS: 17
217
+ POOLER_RESOLUTION: 14
218
+ POOLER_SAMPLING_RATIO: 0
219
+ POOLER_TYPE: ROIAlignV2
220
+ ROI_MASK_HEAD:
221
+ CLS_AGNOSTIC_MASK: false
222
+ CONV_DIM: 256
223
+ NAME: MaskRCNNConvUpsampleHead
224
+ NORM: ''
225
+ NUM_CONV: 4
226
+ POOLER_RESOLUTION: 14
227
+ POOLER_SAMPLING_RATIO: 0
228
+ POOLER_TYPE: ROIAlignV2
229
+ RPN:
230
+ BATCH_SIZE_PER_IMAGE: 256
231
+ BBOX_REG_WEIGHTS:
232
+ - 1.0
233
+ - 1.0
234
+ - 1.0
235
+ - 1.0
236
+ BOUNDARY_THRESH: -1
237
+ HEAD_NAME: StandardRPNHead
238
+ IN_FEATURES:
239
+ - p2
240
+ - p3
241
+ - p4
242
+ - p5
243
+ - p6
244
+ IOU_LABELS:
245
+ - 0
246
+ - -1
247
+ - 1
248
+ IOU_THRESHOLDS:
249
+ - 0.3
250
+ - 0.7
251
+ LOSS_WEIGHT: 1.0
252
+ NMS_THRESH: 0.7
253
+ POSITIVE_FRACTION: 0.5
254
+ POST_NMS_TOPK_TEST: 1000
255
+ POST_NMS_TOPK_TRAIN: 1000
256
+ PRE_NMS_TOPK_TEST: 1000
257
+ PRE_NMS_TOPK_TRAIN: 2000
258
+ SMOOTH_L1_BETA: 0.0
259
+ SEM_SEG_HEAD:
260
+ COMMON_STRIDE: 4
261
+ CONVS_DIM: 128
262
+ IGNORE_VALUE: 255
263
+ IN_FEATURES:
264
+ - p2
265
+ - p3
266
+ - p4
267
+ - p5
268
+ LOSS_WEIGHT: 1.0
269
+ NAME: SemSegFPNHead
270
+ NORM: GN
271
+ NUM_CLASSES: 54
272
+ WEIGHTS: https://www.dropbox.com/s/57zjbwv6gh3srry/model_final.pth?dl=1
273
+ OUTPUT_DIR: ../outputs/publaynet/mask_rcnn_X_101_32x8d_FPN_3x/
274
+ SEED: -1
275
+ SOLVER:
276
+ BASE_LR: 0.01
277
+ BIAS_LR_FACTOR: 1.0
278
+ CHECKPOINT_PERIOD: 5000
279
+ GAMMA: 0.1
280
+ IMS_PER_BATCH: 12
281
+ LR_SCHEDULER_NAME: WarmupMultiStepLR
282
+ MAX_ITER: 20000
283
+ MOMENTUM: 0.9
284
+ STEPS:
285
+ - 7000
286
+ - 12000
287
+ - 15000
288
+ - 17000
289
+ WARMUP_FACTOR: 0.001
290
+ WARMUP_ITERS: 700
291
+ WARMUP_METHOD: linear
292
+ WEIGHT_DECAY: 0.0001
293
+ WEIGHT_DECAY_BIAS: 0.0001
294
+ WEIGHT_DECAY_NORM: 0.0
295
+ TEST:
296
+ AUG:
297
+ ENABLED: false
298
+ FLIP: false
299
+ MAX_SIZE: 4000
300
+ MIN_SIZES:
301
+ - 100
302
+ - 200
303
+ - 400
304
+ - 500
305
+ - 600
306
+ - 700
307
+ - 800
308
+ - 900
309
+ - 1000
310
+ - 1100
311
+ - 1200
312
+ DETECTIONS_PER_IMAGE: 60
313
+ EVAL_PERIOD: 0
314
+ EXPECTED_RESULTS: []
315
+ KEYPOINT_OKS_SIGMAS: []
316
+ PRECISE_BN:
317
+ ENABLED: false
318
+ NUM_ITER: 200
319
+ VERSION: 2
320
+ VIS_PERIOD: 0