-
Notifications
You must be signed in to change notification settings - Fork 1
/
train_net.py
392 lines (333 loc) · 14.5 KB
/
train_net.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
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
# Copyright (c) Facebook, Inc. and its affiliates.
import logging
import os
import sys
from collections import OrderedDict
import torch
from torch.nn.parallel import DistributedDataParallel
import time
import datetime
from fvcore.common.timer import Timer
import detectron2.utils.comm as comm
from detectron2.checkpoint import DetectionCheckpointer#, PeriodicCheckpointer
from detectron2.utils.checkpointing import PeriodicCheckpointer
from detectron2.config import get_cfg
from detectron2.data import (
MetadataCatalog,
build_detection_test_loader,
)
from detectron2.engine import default_argument_parser, default_setup, launch
from detectron2.evaluation import (
inference_on_dataset,
inference_on_dataset_custom,
print_csv_format,
LVISEvaluator,
COCOEvaluator,
)
from detectron2.modeling import build_model
from detectron2.solver import build_lr_scheduler, build_optimizer
from detectron2.utils.events import (
CommonMetricPrinter,
EventStorage,
JSONWriter,
TensorboardXWriter,
)
from detectron2.data.dataset_mapper import DatasetMapper
from detectron2.data.build import build_detection_train_loader, build_detection_val_loader
from detectron2.utils.logger import setup_logger
from torch.cuda.amp import GradScaler
import json
import torch
import numpy as np
import sys
import clip
from collections import Counter
sys.path.insert(0, 'third_party/CenterNet2/')
from centernet.config import add_centernet_config
sys.path.insert(0, 'third_party/Deformable-DETR')
from detic.config import add_detic_config
from detic.data.custom_build_augmentation import build_custom_augmentation
from detic.data.custom_dataset_dataloader import build_custom_train_loader
from detic.data.custom_dataset_mapper import CustomDatasetMapper, DetrDatasetMapper
from detic.custom_solver import build_custom_optimizer
from detic.evaluation.oideval import OIDEvaluator
from detic.evaluation.custom_coco_eval import CustomCOCOEvaluator
from detic.modeling.utils import reset_cls_test#, reset_cls_train
logger = logging.getLogger("detectron2")
def do_test(cfg, model, path_suffix=None):
results = OrderedDict()
for d, dataset_name in enumerate(cfg.DATASETS.TEST):
custom_vocab_bool = len(cfg.DATASETS.ALL_CLASSES)>cfg.MODEL.TEST_NUM_CLASSES[d]
if cfg.MODEL.RESET_CLS_TESTS:
reset_cls_test(
model,
cfg.MODEL.TEST_CLASSIFIERS[d],
cfg.MODEL.TEST_NUM_CLASSES[d])
mapper = None if cfg.INPUT.TEST_INPUT_TYPE == 'default' \
else DatasetMapper(
cfg, False, augmentations=build_custom_augmentation(cfg, False))
data_loader = build_detection_test_loader(cfg, dataset_name, mapper=mapper) # mapper is assigned if it is None, just follow path trace of this function to _test_loader_from_config() in build.py
if path_suffix is not None:
output_folder = os.path.join(
cfg.OUTPUT_DIR, path_suffix, "inference_{}".format(dataset_name))
else:
output_folder = os.path.join(
cfg.OUTPUT_DIR, "inference_{}".format(dataset_name))
evaluator_type = MetadataCatalog.get(dataset_name).evaluator_type
if evaluator_type == "lvis" or cfg.GEN_PSEDO_LABELS:
evaluator = LVISEvaluator(dataset_name, cfg, True, output_folder)
elif evaluator_type == 'coco':
if dataset_name == 'coco_generalized_zeroshot_val':
# Additionally plot mAP for 'seen classes' and 'unseen classes'
evaluator = CustomCOCOEvaluator(dataset_name, cfg, True, output_folder)
else:
evaluator = COCOEvaluator(dataset_name, cfg, True, output_folder)
elif evaluator_type == 'oid':
evaluator = OIDEvaluator(dataset_name, cfg, True, output_folder)
else:
assert 0, evaluator_type
results[dataset_name] = inference_on_dataset_custom(
model, data_loader, evaluator, cfg)
if comm.is_main_process():
logger.info("Evaluation results for {} in csv format:".format(
dataset_name))
print_csv_format(results[dataset_name])
if len(results) == 1:
results = list(results.values())[0]
return results
def do_validation(cfg, data_loader, model, writers=None, train_iteration=-1):
start_iter = 0
print('Running Validation')
with EventStorage(train_iteration) as storage:
step_timer = Timer()
data_timer = Timer()
start_time = time.perf_counter()
total_loss = 0
total_loss_dict = None
data_time = data_timer.seconds()
for iteration, data in enumerate(data_loader):
print(f'Running val iteration {iteration}')
loss_dict = model(data, valmode=True)
loss_dict = {'val_'+str(k):v for k, v in loss_dict.items()}
losses = sum(
loss for k, loss in loss_dict.items())
assert torch.isfinite(losses).all(), loss_dict
loss_dict_reduced = {k: v.item() \
for k, v in comm.reduce_dict(loss_dict).items()}
if total_loss_dict is None:
total_loss_dict = loss_dict_reduced
else:
total_loss_dict = dict(Counter(total_loss_dict) + Counter(loss_dict_reduced))
losses_reduced = sum(loss for loss in loss_dict_reduced.values())
total_loss += losses_reduced
avg_total_loss = total_loss / (iteration+1)
avg_total_loss_dict = { k: v/(iteration+1) for k, v in total_loss_dict.items()}
storage.put_scalars(val_data_time=data_time)
if comm.is_main_process():
storage.put_scalars(
val_total_loss=avg_total_loss, **avg_total_loss_dict)
for writer in writers:
writer.write()
data_timer.reset()
total_time = time.perf_counter() - start_time
logger.info(
"Total training time: {}".format(
str(datetime.timedelta(seconds=int(total_time)))))
return avg_total_loss
def do_train(cfg, model, resume=False):
model.train()
if cfg.SOLVER.USE_CUSTOM_SOLVER:
optimizer = build_custom_optimizer(cfg, model)
else:
assert cfg.SOLVER.OPTIMIZER == 'SGD'
assert cfg.SOLVER.CLIP_GRADIENTS.CLIP_TYPE != 'full_model'
assert cfg.SOLVER.BACKBONE_MULTIPLIER == 1.
optimizer = build_optimizer(cfg, model)
scheduler = build_lr_scheduler(cfg, optimizer)
checkpointer = DetectionCheckpointer(
model, cfg.OUTPUT_DIR, optimizer=optimizer, scheduler=scheduler
)
start_iter = checkpointer.resume_or_load(
cfg.MODEL.WEIGHTS, resume=resume).get("iteration", -1) + 1
if not resume:
start_iter = 0
max_iter = cfg.SOLVER.MAX_ITER if cfg.SOLVER.TRAIN_ITER < 0 else cfg.SOLVER.TRAIN_ITER
periodic_checkpointer = PeriodicCheckpointer(
checkpointer, cfg.SOLVER.CHECKPOINT_PERIOD, max_iter=max_iter
)
periodic_checkpointer_best_ckpt = PeriodicCheckpointer(
checkpointer, cfg.SOLVER.CHECKPOINT_PERIOD, max_iter=max_iter, file_prefix = "best_model", best_model_mode = True
)
writers = (
[
CommonMetricPrinter(max_iter),
JSONWriter(os.path.join(cfg.OUTPUT_DIR, "metrics.json")),
TensorboardXWriter(cfg.OUTPUT_DIR),
]
if comm.is_main_process()
else []
)
val_writers = (
[
CommonMetricPrinter(max_iter),
JSONWriter(os.path.join(cfg.OUTPUT_DIR, "val_metrics.json")),
TensorboardXWriter(cfg.OUTPUT_DIR),
]
if comm.is_main_process()
else []
)
use_custom_mapper = cfg.WITH_IMAGE_LABELS
MapperClass = CustomDatasetMapper if use_custom_mapper else DatasetMapper
mapper = MapperClass(cfg, True, augmentations=[]) if cfg.INPUT.CUSTOM_AUG == '' else \
DetrDatasetMapper(cfg, True) if cfg.INPUT.CUSTOM_AUG == 'DETR' else \
MapperClass(cfg, True, augmentations=build_custom_augmentation(cfg, True))
if cfg.DATALOADER.SAMPLER_TRAIN in ['TrainingSampler', 'RepeatFactorTrainingSampler']:
data_loader = build_detection_train_loader(cfg, mapper=mapper)
else:
data_loader = build_custom_train_loader(cfg, mapper=mapper)
if cfg.FP16:
scaler = GradScaler()
if cfg.TEST.VAL_PERIOD > 0:
mapper = None if cfg.INPUT.TEST_INPUT_TYPE == 'default' \
else DatasetMapper( # check val_loader_from_config() to edit augmentations in Mapper. The issue is that when mapper is None, augmentations were applied by default. Fixed those now with empty lists.
cfg, False, augmentations=build_custom_augmentation(cfg, False))
if cfg.DATALOADER.SAMPLER_TRAIN in ['TrainingSampler', 'RepeatFactorTrainingSampler', 'InferenceSampler']:
val_data_loader = build_detection_val_loader(cfg, mapper=mapper)
logger.info("Starting training from iteration {}".format(start_iter))
with EventStorage(start_iter) as storage:
step_timer = Timer()
data_timer = Timer()
start_time = time.perf_counter()
total_val_loss_dict = None
best_val_loss = 99999
for data, iteration in zip(data_loader, range(start_iter, max_iter)):
data_time = data_timer.seconds()
storage.put_scalars(data_time=data_time)
step_timer.reset()
iteration = iteration + 1
storage.step()
loss_dict = model(data, valmode=False)
losses = sum(
loss for k, loss in loss_dict.items())
assert torch.isfinite(losses).all(), loss_dict
loss_dict_reduced = {k: v.item() \
for k, v in comm.reduce_dict(loss_dict).items()}
losses_reduced = sum(loss for loss in loss_dict_reduced.values())
if comm.is_main_process():
storage.put_scalars(
total_loss=losses_reduced, **loss_dict_reduced)
optimizer.zero_grad()
if cfg.FP16:
scaler.scale(losses).backward()
scaler.step(optimizer)
scaler.update()
else:
losses.backward()
optimizer.step()
storage.put_scalar(
"lr", optimizer.param_groups[0]["lr"], smoothing_hint=False)
step_time = step_timer.seconds()
storage.put_scalars(time=step_time)
data_timer.reset()
scheduler.step()
if (cfg.TEST.VAL_PERIOD > 0
and iteration % cfg.TEST.VAL_PERIOD == 0):
cur_val_loss = do_validation(cfg, val_data_loader, model, val_writers, train_iteration=iteration)
if cur_val_loss < best_val_loss:
best_val_loss = cur_val_loss
best_iteration = iteration
periodic_checkpointer_best_ckpt.step(iteration)
comm.synchronize()
if iteration - start_iter > 5 and \
(iteration % 20 == 0 or iteration == max_iter):# and not iteration % cfg.TEST.VAL_PERIOD != 0:
for writer in writers:
writer.write()
periodic_checkpointer.step(iteration)
total_time = time.perf_counter() - start_time
logger.info(
"Total training time: {}".format(
str(datetime.timedelta(seconds=int(total_time)))))
if cfg.TEST.VAL_PERIOD > 0:
old_best_model_name = os.path.join(cfg.OUTPUT_DIR, "best_model_final.pth")
new_best_model_name = os.path.join(cfg.OUTPUT_DIR, f"best_model_final_{best_iteration}.pth")
os.rename(old_best_model_name, new_best_model_name)
return new_best_model_name
else:
return None
def setup(args):
"""
Create configs and perform basic setups.
"""
cfg = get_cfg()
add_centernet_config(cfg)
add_detic_config(cfg)
cfg.merge_from_file(args.config_file)
cfg.merge_from_list(args.opts)
if cfg.OUTPUT_DIR_PREFIX is not None:
suffix_file_name = args.config_file.replace('configs/my_configs/','')[:-5]
cfg.OUTPUT_DIR = os.path.join(cfg.OUTPUT_DIR_PREFIX, suffix_file_name)
if not args.pred_all_class:
cfg.MODEL.ROI_HEADS.ONE_CLASS_PER_PROPOSAL = True
cfg.freeze()
default_setup(cfg, args)
setup_logger(output=cfg.OUTPUT_DIR, \
distributed_rank=comm.get_rank(), name="detic")
return cfg
def main(args):
cfg = setup(args)
model = build_model(cfg)
logger.info("Model:\n{}".format(model))
if args.eval_only:
DetectionCheckpointer(model, save_dir=cfg.OUTPUT_DIR).resume_or_load(
cfg.MODEL.WEIGHTS, resume=args.resume
)
return do_test(cfg, model)
else:
if cfg.MODEL.RESET_CLS_TRAIN:
reset_cls_test(
model,
cfg.MODEL.TRAIN_CLASSIFIERS[0],
cfg.MODEL.TRAIN_NUM_CLASSES[0])
[]
distributed = comm.get_world_size() > 1
if distributed:
model = DistributedDataParallel(
model, device_ids=[comm.get_local_rank()], broadcast_buffers=False,
find_unused_parameters=cfg.FIND_UNUSED_PARAM
)
best_model_path = do_train(cfg, model, resume=args.resume)
if best_model_path is not None: # i.e using valset
DetectionCheckpointer(model, save_dir=cfg.OUTPUT_DIR).resume_or_load(
best_model_path, resume=args.resume
)
print('Best model Path', best_model_path)
print('Running Test script with best model weights')
return do_test(cfg, model, path_suffix=best_model_path.split('.pth')[0])
else:
return []
if __name__ == "__main__":
args = default_argument_parser()
args = args.parse_args()
if args.num_machines == 1:
args.dist_url = 'tcp://127.0.0.1:{}'.format(
torch.randint(11111, 60000, (1,))[0].item())
else:
if args.dist_url == 'host':
args.dist_url = 'tcp://{}:12345'.format(
os.environ['SLURM_JOB_NODELIST'])
elif not args.dist_url.startswith('tcp'):
tmp = os.popen(
'echo $(scontrol show job {} | grep BatchHost)'.format(
args.dist_url)
).read()
tmp = tmp[tmp.find('=') + 1: -1]
args.dist_url = 'tcp://{}:12345'.format(tmp)
print("Command Line Args:", args)
launch(
main,
args.num_gpus,
num_machines=args.num_machines,
machine_rank=args.machine_rank,
dist_url=args.dist_url,
args=(args,),
)