-
Notifications
You must be signed in to change notification settings - Fork 23
/
test.py
178 lines (157 loc) · 7.93 KB
/
test.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
import argparse
import os
import numpy as np
import time
from modeling.coanet import *
from dataloaders import custom_transforms as tr
from PIL import Image
from torchvision import transforms
from dataloaders.utils import *
from torchvision.utils import make_grid #, save_image
from dataloaders import make_data_loader
from utils.metrics import Evaluator
from utils.loss import SegmentationLosses
from tqdm import tqdm
os.environ['CUDA_VISIBLE_DEVICES'] = '0'
def save_image(tensor, filename, nrow=8, padding=2,
normalize=False, range=None, scale_each=False, pad_value=0):
"""Save a given Tensor into an image file.
Args:
tensor (Tensor or list): Image to be saved. If given a mini-batch tensor,
saves the tensor as a grid of images by calling ``make_grid``.
**kwargs: Other arguments are documented in ``make_grid``.
"""
from PIL import Image
grid = make_grid(tensor, nrow=nrow, padding=padding, pad_value=pad_value,
normalize=normalize, range=range, scale_each=scale_each)
# Add 0.5 after unnormalizing to [0, 255] to round to nearest integer
ndarr = grid.mul_(255).add_(0.5).clamp_(0, 255).permute(1, 2, 0).to('cpu', torch.uint8).numpy()
im = Image.fromarray(ndarr)
im = im.resize([1300, 1300])
im.save(filename)
def main():
parser = argparse.ArgumentParser(description="PyTorch CoANet Training")
parser.add_argument('--out-path', type=str, default='./run/spacenet/CoANet-resnet',
help='mask image to save')
parser.add_argument('--backbone', type=str, default='resnet',
help='backbone name (default: resnet)')
parser.add_argument('--batch-size', type=int, default=1,
metavar='N', help='input batch size for test ')
parser.add_argument('--ckpt', type=str, default='./run/spacenet/CoANet-resnet/CoANet-spacenet.pth.tar',
help='saved model')
parser.add_argument('--out-stride', type=int, default=8,
help='network output stride (default: 8)')
parser.add_argument('--loss-type', type=str, default='con_ce',
choices=['ce', 'con_ce', 'focal'],
help='loss func type')
parser.add_argument('--workers', type=int, default=16,
metavar='N', help='dataloader threads')
parser.add_argument('--no-cuda', action='store_true', default=False,
help='disables CUDA training')
parser.add_argument('--gpu-ids', type=str, default='0',
help='use which gpu to train, must be a \
comma-separated list of integers only (default=0)')
parser.add_argument('--dataset', type=str, default='spacenet',
choices=['spacenet', 'DeepGlobe'],
help='dataset name')
parser.add_argument('--base-size', type=int, default=1280,
help='base image size. spacenet:1280, DeepGlobe:1024.')
parser.add_argument('--crop-size', type=int, default=1280,
help='crop image size. spacenet:1280, DeepGlobe:1024.')
parser.add_argument('--sync-bn', type=bool, default=False,
help='whether to use sync bn')
parser.add_argument('--freeze-bn', type=bool, default=False,
help='whether to freeze bn parameters (default: False)')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
if args.cuda:
try:
args.gpu_ids = [int(s) for s in args.gpu_ids.split(',')]
except ValueError:
raise ValueError('Argument --gpu_ids must be a comma-separated list of integers only')
kwargs = {'num_workers': args.workers, 'pin_memory': False}
train_loader, val_loader, test_loader, nclass = make_data_loader(args, **kwargs)
model = CoANet(num_classes=nclass,
backbone=args.backbone,
output_stride=args.out_stride,
sync_bn=args.sync_bn,
freeze_bn=args.freeze_bn)
model = model.cuda()
ckpt = torch.load(args.ckpt)
model.load_state_dict(ckpt['state_dict'])
out_path = os.path.join(args.out_path, 'out_imgs_1300/')
if not os.path.exists(out_path):
os.makedirs(out_path)
evaluator = Evaluator(2)
model.eval()
evaluator.reset()
tbar = tqdm(test_loader, desc='\r')
for i, sample in enumerate(tbar):
image, target = sample[0]['image'], sample[0]['label']
image = image.cpu().numpy()
image1 = image[:, :, ::-1, :]
image2 = image[:, :, :, ::-1]
image3 = image[:, :, ::-1, ::-1]
image = np.concatenate((image,image1,image2,image3), axis=0)
image = torch.from_numpy(image).float()
img_name = sample[1][0].split('.')[0]
if args.cuda:
image, target = image.cuda(), target.cuda()
with torch.no_grad():
output, out_connect, out_connect_d1 = model(image)
out_connect_full = []
out_connect = out_connect.data.cpu().numpy()
out_connect_full.append(out_connect[0, ...])
out_connect_full.append(out_connect[1, :, ::-1, :])
out_connect_full.append(out_connect[2, :, :, ::-1])
out_connect_full.append(out_connect[3, :, ::-1, ::-1])
out_connect_full = np.asarray(out_connect_full).mean(axis=0)[np.newaxis, :, :, :]
pred_connect = np.sum(out_connect_full, axis=1)
pred_connect[pred_connect < 0.9] = 0
pred_connect[pred_connect >= 0.9] = 1
out_connect_d1_full = []
out_connect_d1 = out_connect_d1.data.cpu().numpy()
out_connect_d1_full.append(out_connect_d1[0, ...])
out_connect_d1_full.append(out_connect_d1[1, :, ::-1, :])
out_connect_d1_full.append(out_connect_d1[2, :, :, ::-1])
out_connect_d1_full.append(out_connect_d1[3, :, ::-1, ::-1])
out_connect_d1_full = np.asarray(out_connect_d1_full).mean(axis=0)[np.newaxis, :, :, :]
pred_connect_d1 = np.sum(out_connect_d1_full, axis=1)
pred_connect_d1[pred_connect_d1 < 2.0] = 0
pred_connect_d1[pred_connect_d1 >= 2.0] = 1
pred_full = []
pred = output.data.cpu().numpy()
target_n = target.cpu().numpy()
pred_full.append(pred[0, ...])
pred_full.append(pred[1, :, ::-1, :])
pred_full.append(pred[2, :, :, ::-1])
pred_full.append(pred[3, :, ::-1, ::-1])
pred_full = np.asarray(pred_full).mean(axis=0)
pred_full[pred_full > 0.1] = 1
pred_full[pred_full < 0.1] = 0
su = pred_full + pred_connect + pred_connect_d1
su[su > 0] = 1
evaluator.add_batch(target_n, su.astype(int))#
# save imgs
out_image = make_grid(image[0,:].clone().cpu().data, 3, normalize=True)
out_GT = make_grid(decode_seg_map_sequence(torch.squeeze(target[:3], 1).detach().cpu().numpy(),
dataset=args.dataset), 3, normalize=False, range=(0, 255))
out_pred_label_sum = make_grid(decode_seg_map_sequence(su,
dataset=args.dataset), 3, normalize=False, range=(0, 255))
save_image(out_image, out_path + img_name + '_sat.png')
save_image(out_GT, out_path + img_name + '_GT' + '.png')
save_image(out_pred_label_sum, out_path + img_name + '_pred' + '.png')
# Fast test during the training
Acc = evaluator.Pixel_Accuracy()
Acc_class = evaluator.Pixel_Accuracy_Class()
mIoU = evaluator.Mean_Intersection_over_Union()
IoU = evaluator.Intersection_over_Union()
Precision = evaluator.Pixel_Precision()
Recall = evaluator.Pixel_Recall()
F1 = evaluator.Pixel_F1()
print('Validation:')
print('[numImages: %5d]' % (i * args.batch_size + image.data.shape[0]))
print("Acc:{}, Acc_class:{}, mIoU:{}, IoU:{}, Precision:{}, Recall:{}, F1:{}"
.format(Acc, Acc_class, mIoU, IoU, Precision, Recall, F1))
if __name__ == "__main__":
main()