-
Notifications
You must be signed in to change notification settings - Fork 3
/
regressors_main.py
127 lines (108 loc) · 3.29 KB
/
regressors_main.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
import os
from PIL import Image
import torch
import torch.nn as nn
from torch.utils.data import DataLoader
from torchvision import transforms
from torch.optim import lr_scheduler
from torch.utils.tensorboard import SummaryWriter
from Datasets.Morph2.DataParser import DataParser
from Datasets.AFAD.AFADRegressorDataset import AFADRegressorDataset
from Training.train_regression_model import train_regression_model
from Models.AgeMultiHeadRegressor import AgeMultiHeadRegressor
from Optimizers.RangerLars import RangerLars
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
print(device)
torch.cuda.empty_cache()
age_interval = 5
min_age = 15
max_age = 40
BATCH_SIZE = 128
num_labels = int(max_age / age_interval - min_age / age_interval + 1)
# Load data
# data_parser = DataParser('./Datasets/Morph2/aligned_data/aligned_dataset_with_metadata_uint8.hdf5')
# data_parser.initialize_data()
#
# train_ds = Morph2RegressorDataset(
# data_parser.x_train,
# data_parser.y_train,
# min_age,
# age_intareval,
# num_labels,
# transform=transforms.Compose([
# transforms.RandomResizedCrop(160, (0.95, 1.0)),
# transforms.RandomHorizontalFlip(),
# transforms.ToTensor()
# ])
# )
#
# test_ds = Morph2RegressorDataset(
# data_parser.x_test,
# data_parser.y_test,
# min_age,
# age_intareval,
# num_labels,
# transform=transforms.Compose([
# transforms.ToTensor()
# ])
# )
train_ds = AFADRegressorDataset(
'./Datasets/AFAD/aligned_data/afad_train.h5',
min_age=min_age,
max_age=max_age,
age_interval=age_interval,
transform=transforms.Compose([
# transforms.Normalize((103.939, 116.779, 123.68), (1, 1, 1)),
transforms.RandomResizedCrop(160, (0.9, 1.0)),
transforms.RandomHorizontalFlip(),
transforms.RandomRotation(degrees=10, resample=Image.BICUBIC),
transforms.ToTensor()
])
)
test_ds = AFADRegressorDataset(
'./Datasets/AFAD/aligned_data/afad_test.h5',
min_age=min_age,
max_age=max_age,
age_interval=age_interval,
transform=transforms.Compose([
transforms.ToTensor()
])
)
image_datasets = {
'train': train_ds,
'val': test_ds
}
data_loaders = {
x: DataLoader(image_datasets[x], batch_size=BATCH_SIZE, shuffle=True, num_workers=0)
for x in ['train', 'val']
}
dataset_sizes = {x: len(image_datasets[x]) for x in ['train', 'val']}
# create model and parameters
multihead_regression_model = AgeMultiHeadRegressor(num_labels, age_interval, min_age, max_age)
multihead_regression_model.to(device)
multihead_regression_model.freeze_base_cnn(True)
criterion = nn.MSELoss().to(device)
optimizer = RangerLars(multihead_regression_model.parameters(), lr=1e-2)
exp_lr_scheduler = lr_scheduler.StepLR(optimizer, step_size=10, gamma=0.1)
### Train ###
writer = SummaryWriter(
'logs/AFAD/multihead_regression/RangerLars_unfreeze_at_15_lr_1e2_steplr_10_01_256_dropout')
best_classification_model = train_regression_model(
multihead_regression_model,
criterion,
optimizer,
exp_lr_scheduler,
data_loaders,
dataset_sizes,
device,
writer,
num_labels,
num_epochs=30
)
print('saving best model')
model_path = 'weights/AFAD/multihead_regression/RangerLars_unfreeze_at_15_lr_1e2_steplr_10_01_256_dropout'
if not os.path.exists(model_path):
os.makedirs(model_path)
FINAL_MODEL_FILE = os.path.join(model_path, "weights.pt")
torch.save(best_classification_model.state_dict(), FINAL_MODEL_FILE)
print('exiting')