-
Notifications
You must be signed in to change notification settings - Fork 0
/
main_voc.py
175 lines (156 loc) · 7.12 KB
/
main_voc.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
import argparse
import os
import time
import cv2
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from torch.utils.data import Dataset
import torchvision.models as models
from torch.autograd import Variable
import numpy as np
from tensorboardX import SummaryWriter
from augmentations import SSDAugmentation
from voc0712 import *
from coco import *
from model import CDMP_Localization
parser = argparse.ArgumentParser(description='cdmp-localization')
parser.add_argument('--tag', type=str, default='default')
parser.add_argument('--epoch', type=int, default=200)
parser.add_argument('--mode', choices=['train', 'test'], required=True)
parser.add_argument('--batch-size', type=int, default=2)
parser.add_argument('--cuda', action='store_true')
parser.add_argument('--lr', type=float, default=0.01)
parser.add_argument('--model-path', type=str, default='./assets/learned_models',
help='pre-trained model path')
parser.add_argument('--log-interval', type=int, default=10)
parser.add_argument('--save-interval', type=int, default=10)
args = parser.parse_args()
args.cuda = args.cuda if torch.cuda.is_available else False
if args.cuda:
torch.cuda.manual_seed(1)
logger = SummaryWriter(os.path.join('./assets/log/', args.tag))
np.random.seed(int(time.time()))
device_id = [0,1,2]
# dataset hyper params
min_dim = 224
object_size = 224
means = (104, 117, 123)
train_loader = torch.utils.data.DataLoader(
VOCLocalization(root='/home1/mxj/workspace/dataset/voc/VOCdevkit',
transform=SSDAugmentation(min_dim, means),
image_sets=[('2012', 'trainval')],
object_size=object_size,
),
batch_size=args.batch_size,
num_workers=64,
pin_memory=True,
shuffle=True,
collate_fn=localization_collate
)
test_loader = torch.utils.data.DataLoader(
VOCLocalization(root='/home1/mxj/workspace/dataset/voc/VOCdevkit',
transform=SSDAugmentation(min_dim, means),
image_sets=[('2007', 'trainval')],
object_size=object_size,
),
batch_size=args.batch_size,
num_workers=64,
pin_memory=True,
shuffle=True,
collate_fn=localization_collate
)
def train(model, loader, epoch, optimizer):
model.train()
for batch_idx, (img, object_img, target) in enumerate(loader):
# img: (N, C, H, W)
# object_img: (N, C, H, W)
# target: (N, 3) [x, y, id]
if args.cuda:
img, object_img, target = img.cuda(), object_img.cuda(), target.cuda()
img, object_img, target = Variable(img), Variable(object_img), Variable(target)
optimizer.zero_grad()
output = model(img, object_img)
loss = F.mse_loss(output, target[:, :2]).mean() # because you've already using log_softmax as output
loss.backward()
optimizer.step()
if batch_idx % args.log_interval == 0:
print('Train Epoch: {} [{}/{} ({:.0f}%)]\tLoss: {:.6f}'.format(
epoch, batch_idx * args.batch_size, len(loader.dataset),
100. * batch_idx * args.batch_size / len(loader.dataset), loss.data[0]))
logger.add_scalar('train_loss', loss.cpu().data[0]/args.batch_size,
batch_idx + epoch * len(loader))
# visualize gt
x, y = (output[0]*min_dim).data.cpu().numpy().astype(np.int32)
gt_x, gt_y = (target[0]*min_dim).data.cpu().numpy().astype(np.int32)[:2]
label_id = int(target[0, -1])
log_img = img[0].permute(1,2,0).data.cpu().numpy().astype(np.uint8)
log_img += np.array(means).astype(np.uint8)
log_img = cv2.circle(log_img, (x, y), 10, (255,0,0), 5)
cv2.putText(log_img, VOC_CLASSES[label_id], (x, y), cv2.FONT_HERSHEY_SIMPLEX, 0.3,
(255,0,0))
log_img = cv2.circle(log_img, (gt_x, gt_y), 10, (0,255,0), 5)
cv2.putText(log_img, VOC_CLASSES[label_id], (gt_x, gt_y), cv2.FONT_HERSHEY_SIMPLEX, 0.3,
(0,255,0))
obj_img = object_img[0].permute(1,2,0).data.cpu().numpy().astype(np.uint8)
obj_img += np.array(means).astype(np.uint8)
logger.add_image('train_log_img', log_img, batch_idx + epoch * len(loader))
logger.add_image('train_obj_img', obj_img, batch_idx + epoch * len(loader))
def test(model, loader, epoch):
model.eval()
test_loss = 0
for batch_idx, (img, object_img, target) in enumerate(loader):
# img: (N, C, H, W)
# object_img: (N, C, H, W)
# target: (N, 3) [x, y, id]
if args.cuda:
img, object_img, target = img.cuda(), object_img.cuda(), target.cuda()
img, object_img, target = Variable(img), Variable(object_img), Variable(target)
output = model(img, object_img)
loss = F.mse_loss(output, target[:, :2]).mean() # because you've already using log_softmax as output
test_loss += loss.data.cpu()
if batch_idx % args.log_interval == 0:
# visualize gt
x, y = (output[0]*min_dim).data.cpu().numpy().astype(np.int32)
gt_x, gt_y, label_id = (target[0]*min_dim).data.cpu().numpy().astype(np.int32)
label_id = int(target[0, -1])
log_img = img[0].permute(1,2,0).data.cpu().numpy().astype(np.uint8)
log_img = cv2.circle(log_img, (x, y), 10, (255,0,0), 5)
log_img += np.array(means).astype(np.uint8)
cv2.putText(log_img, VOC_CLASSES[label_id], (x, y), cv2.FONT_HERSHEY_SIMPLEX, 0.3,
(255,0,0))
log_img = cv2.circle(log_img, (gt_x, gt_y), 10, (0,255,0), 5)
cv2.putText(log_img, VOC_CLASSES[label_id], (gt_x, gt_y), cv2.FONT_HERSHEY_SIMPLEX, 0.3,
(0,255,0))
obj_img = object_img[0].permute(1,2,0).data.cpu().numpy().astype(np.uint8)
obj_img += np.array(means).astype(np.uint8)
logger.add_image('test_log_img', log_img, batch_idx + epoch * len(loader))
logger.add_image('test_obj_img', obj_img, batch_idx + epoch * len(loader))
test_loss /= len(loader.dataset)
# visualize gt
# TBD
return test_loss
def main():
if args.mode == 'train':
model = CDMP_Localization(input_size=min_dim, object_size=object_size)
if args.cuda:
model = nn.DataParallel(model, device_ids=device_id).cuda()
# model = model.cuda()
optimizer = optim.Adam(model.parameters(), lr=args.lr)
for epoch in range(args.epoch):
train(model, train_loader, epoch, optimizer)
loss = test(model, test_loader, epoch)
logger.add_scalar('test_loss', loss, epoch)
if epoch % args.save_interval == 0:
torch.save(model, os.path.join(args.model_path,
args.tag + '_{}.model'.format(epoch)))
else:
model = torch.load(os.path.join(args.model_path, args.tag + '.model'))
if args.cuda:
model = nn.DataParallel(model, device_ids=device_id).cuda()
# model = model.cuda()
loss = test(model, test_loader, 0)
print('Test done, loss={}'.format(loss))
if __name__ == "__main__":
main()