-
Notifications
You must be signed in to change notification settings - Fork 420
/
train_pytorch.py
288 lines (254 loc) · 12.1 KB
/
train_pytorch.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
# Copyright (C) 2021, Mindee.
# This program is licensed under the Apache License version 2.
# See LICENSE or go to <https://www.apache.org/licenses/LICENSE-2.0.txt> for full license details.
import os
os.environ['USE_TORCH'] = '1'
import datetime
import multiprocessing as mp
import logging
import random
import numpy as np
import torch
import torch.optim as optim
import torchvision
import wandb
from fastprogress.fastprogress import master_bar, progress_bar
from torch.utils.data import DataLoader, RandomSampler, SequentialSampler
from torchvision import transforms
from torchvision.ops import MultiScaleRoIAlign
from doctr.datasets import DocArtefacts
from doctr.utils import DetectionMetric
from doctr.transforms.functional.pytorch import rotate
def random_horizontal_flip(img: torch.Tensor, bbox: np.ndarray, width):
if bool(random.getrandbits(1)):
trans = transforms.RandomHorizontalFlip(p=1)
transformed_ = trans(img)
bbox_dup = bbox[:]
for k in range(len(bbox)):
for id in range(len(bbox[k])):
bbox_dup[k][id][0], bbox_dup[k][id][2] = width - bbox[k][id][2], width - bbox[k][id][0]
return transformed_, bbox_dup
else:
return img, bbox
def random_vertical_flip(img: torch.Tensor, bbox: np.ndarray, height):
if bool(random.getrandbits(1)):
trans = transforms.RandomVerticalFlip(p=1.0)
transformed_ = trans(img)
bbox_dup = bbox[:]
for k in range(len(bbox)):
for id in range(len(bbox[k])):
bbox_dup[k][id][3], bbox_dup[k][id][1] = height - bbox[k][id][1], height - bbox[k][id][3]
return (transformed_, bbox_dup)
else:
return img, bbox
def data_augmentations(targets, max_angle, **kwargs):
bbox = np.array([i["boxes"] for i in targets], dtype=object)
images = kwargs.get('images', )
height = kwargs.get('height', )
width = kwargs.get('width', )
angle = 2 * max_angle * (np.random.rand(len(targets)) - 1)
for id in range(len(bbox)):
images[id], bbox[id] = rotate(images[id], bbox[id], angle[id] if bool(
random.getrandbits(1)) else -angle[id])
bbox_t = bbox[id][:, :4]
for z in range(len(bbox[id])):
bbox_t[z][0] = bbox[id][z][0] - bbox[id][z][2] / 2
bbox_t[z][1] = bbox[id][z][1] - bbox[id][z][3] / 2
bbox_t[z][2] = bbox_t[z][0] + bbox[id][z][2]
bbox_t[z][3] = bbox_t[z][1] + bbox[id][z][3]
bbox[id] = bbox_t
images, bbox = random_vertical_flip(images, bbox, height)
images, bbox = random_horizontal_flip(images, bbox, width)
targets = [{
"boxes": torch.from_numpy(bo).to(dtype=torch.float32),
"labels": torch.tensor(t['labels']).to(dtype=torch.long)}
for bo, t in zip(bbox, targets)
]
return images, targets
def convert_to_abs_coords(targets, img_shape, use_aug=False, **kwargs):
height, width = img_shape[-2:]
for idx in range(len(targets)):
targets[idx]['boxes'][:, 0::2] = (targets[idx]['boxes'][:, 0::2] * width).round()
targets[idx]['boxes'][:, 1::2] = (targets[idx]['boxes'][:, 1::2] * height).round()
if use_aug:
images, targets = data_augmentations(targets, 5, **kwargs)
return images, targets
else:
targets = [{
"boxes": torch.from_numpy(t['boxes']).to(dtype=torch.float32),
"labels": torch.tensor(t['labels']).to(dtype=torch.long)}
for t in targets
]
return targets
def fit_one_epoch(model, train_loader, optimizer, scheduler, mb, use_aug=True, ):
model.train()
train_iter = iter(train_loader)
# Iterate over the batches of the dataset
for images, targets in progress_bar(train_iter, parent=mb):
optimizer.zero_grad()
if use_aug:
height, width = images.shape[-2:]
kwargs = {"images": images, "height": height, "width": width, }
images, targets = convert_to_abs_coords(targets, images.shape, use_aug=True, **kwargs)
else:
targets = convert_to_abs_coords(targets, images.shape)
if torch.cuda.is_available():
images = images.cuda()
targets = [{k: v.cuda() for k, v in t.items()} for t in targets]
loss_dict = model(images, targets)
loss = sum(v for v in loss_dict.values())
loss.backward()
optimizer.step()
mb.child.comment = f'Training loss: {loss.item()}'
scheduler.step()
@torch.no_grad()
def evaluate(model, val_loader, metric):
model.eval()
metric.reset()
val_iter = iter(val_loader)
for images, targets in val_iter:
images, targets = next(val_iter)
# batch_transforms
targets = convert_to_abs_coords(targets, images.shape)
if torch.cuda.is_available():
images = images.cuda()
output = model(images)
pred_labels = np.concatenate([o['labels'].cpu().numpy() for o in output])
pred_boxes = np.concatenate([o['boxes'].cpu().numpy() for o in output])
gt_boxes = np.concatenate([o['boxes'].cpu().numpy() for o in targets])
gt_labels = np.concatenate([o['labels'].cpu().numpy() for o in targets])
metric.update(gt_boxes, pred_boxes, gt_labels, pred_labels)
recall, precision, mean_iou = metric.summary()
return recall, precision, mean_iou
def main(args):
torch.backends.cudnn.benchmark = True
if not isinstance(args.workers, int):
args.workers = min(16, mp.cpu_count())
# Filter keys
state_dict = {
k: v for k, v in torchvision.models.detection.__dict__[args.arch](pretrained=True).state_dict().items()
if not k.startswith('roi_heads.')
}
defaults = {"min_size": 700, "max_size": 1300,
"box_fg_iou_thresh": 0.5,
"box_bg_iou_thresh": 0.5,
"box_detections_per_img": 150, "box_score_thresh": 0.15, "box_positive_fraction": 0.35,
"box_nms_thresh": 0.2,
"rpn_pre_nms_top_n_train": 2000, "rpn_pre_nms_top_n_test": 1000,
"rpn_post_nms_top_n_train": 2000, "rpn_post_nms_top_n_test": 1000,
"rpn_nms_thresh": 0.2,
"rpn_batch_size_per_image": 250
}
kwargs = {**defaults}
model = torchvision.models.detection.__dict__[args.arch](pretrained=False, num_classes=5, **kwargs)
model.load_state_dict(state_dict, strict=False)
model.roi_heads.box_roi_pool = MultiScaleRoIAlign(featmap_names=['0', '1', '2', '3'], output_size=(7, 7),
sampling_ratio=2)
anchor_sizes = ((16), (64), (128), (264))
aspect_ratios = ((0.5, 1.0, 2.0, 3.0,)) * len(anchor_sizes)
model.rpn.anchor_generator.sizes = anchor_sizes
model.rpn.anchor_generator.aspect_ratios = aspect_ratios
# GPU
if isinstance(args.device, int):
if not torch.cuda.is_available():
raise AssertionError("PyTorch cannot access your GPU. Please investigate!")
if args.device >= torch.cuda.device_count():
raise ValueError("Invalid device index")
# Silent default switch to GPU if available
elif torch.cuda.is_available():
args.device = 0
else:
logging.warning("No accessible GPU, target device set to CPU.")
if torch.cuda.is_available():
torch.cuda.set_device(args.device)
model = model.cuda()
optimizer = optim.SGD([p for p in model.parameters() if p.requires_grad],
lr=args.lr, weight_decay=args.weight_decay)
scheduler = optim.lr_scheduler.StepLR(optimizer, step_size=5, gamma=0.1)
train_set = DocArtefacts(train=True, download=True)
val_set = DocArtefacts(train=False, download=True)
train_loader = DataLoader(train_set, batch_size=args.batch_size, num_workers=args.workers,
sampler=RandomSampler(train_set), pin_memory=torch.cuda.is_available(),
collate_fn=train_set.collate_fn,
drop_last=True)
val_loader = DataLoader(val_set, batch_size=args.batch_size, num_workers=args.workers,
sampler=SequentialSampler(val_set), pin_memory=torch.cuda.is_available(),
collate_fn=val_set.collate_fn,
drop_last=False)
metric = DetectionMetric(iou_thresh=0.5)
if args.test_only:
print("Running evaluation")
recall, precision, mean_iou = evaluate(model, val_loader, metric)
print(f"Recall: {recall:.2%} | Precision: {precision:.2%} | IoU: {mean_iou:.2%}")
return
# Training monitoring
current_time = datetime.datetime.now().strftime("%Y%m%d-%H%M%S")
exp_name = f"{args.arch}_{current_time}" if args.name is None else args.name
# W&B
if args.wb:
run = wandb.init(
name=exp_name,
project="object-detection",
config={
"learning_rate": args.lr,
"epochs": args.epochs,
"weight_decay": args.weight_decay,
"batch_size": args.batch_size,
"architecture": args.arch,
"input_size": args.input_size,
"optimizer": "sgd",
"framework": "pytorch",
"scheduler": args.sched,
"pretrained": args.pretrained,
}
)
mb = master_bar(range(args.epochs))
max_score = 0.
for epoch in mb:
fit_one_epoch(model, train_loader, optimizer, scheduler, mb, use_aug=args.use_augmentations)
recall, precision, mean_iou = evaluate(model, val_loader, metric)
f1_score = 2 * precision * recall / (precision + recall) if (precision + recall) > 0 else 0.
mb.write(
f"Epoch {epoch + 1}/{args.epochs} - "
f"Recall: {recall:.2%} | Precision: {precision:.2%} "
f"|IoU: {mean_iou:.2%}")
# W&B
if args.wb:
wandb.log({
'recall': recall,
'precision': precision,
'iou': mean_iou,
})
if f1_score > max_score:
print(f"Validation metric increased {max_score:.6} --> {f1_score:.6}: saving state...")
# torch.save(model.state_dict(), f"./{exp_name}.pt")
torch.save(model.state_dict(), f"./{epoch}.pt")
max_score = f1_score
if args.wb:
run.finish()
def parse_args():
import argparse
parser = argparse.ArgumentParser(description='DocTR training script for object detection (PyTorch)',
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
parser.add_argument('arch', type=str, help='text-detection model to train')
parser.add_argument('--name', type=str, default=None, help='Name of your training experiment')
parser.add_argument('--epochs', type=int, default=20, help='number of epochs to train the model on')
parser.add_argument('-b', '--batch_size', type=int, default=8, help='batch size for training')
parser.add_argument('--device', default=None, type=int, help='device')
parser.add_argument('--use_augmentations', default=True, type=bool, help='augmentations')
parser.add_argument('--input_size', type=int, default=1024, help='model input size, H = W')
parser.add_argument('--lr', type=float, default=0.001, help='learning rate for the optimizer (SGD)')
parser.add_argument('--wd', '--weight-decay', default=0, type=float, help='weight decay', dest='weight_decay')
parser.add_argument('-j', '--workers', type=int, default=None, help='number of workers used for dataloading')
parser.add_argument('--resume', type=str, default=None, help='Path to your checkpoint')
parser.add_argument('--wb', dest='wb', action='store_true',
help='Log to Weights & Biases')
parser.add_argument('--pretrained', dest='pretrained', action='store_true',
help='Load pretrained parameters before starting the training')
parser.add_argument('--sched', type=str, default='cosine', help='scheduler to use')
parser.add_argument("--test-only", dest='test_only', action='store_true', help="Run the validation loop")
args = parser.parse_args()
return args
if __name__ == "__main__":
args = parse_args()
main(args)