-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathtest_Ablation.py
253 lines (216 loc) · 12.1 KB
/
test_Ablation.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
import os
#os.environ['CUDA_VISIBLE_DEVICES'] = '0'
import torch
import torch.nn.functional as F
import numpy as np
from utils import data_loader
from tqdm import tqdm
from utils.metrics import Evaluator
# from network.Net import HANet_v2
from PIL import Image
from network.Net2 import HANet_v2,HANet_v3,HANet_v4
from network.Net3 import HANet_v5,HANet_v6,HANet_v6_Ablation
from network.CD_Model import HANModel3
from network.SemiModel import SemiModel
from network import Ablation_model
import time
start=time.time()
def test(test_loader, Eva_test, save_path, net):
print("Strat validing!")
net.train(False)
net.eval()
for i, (A, B, mask, filename) in enumerate(tqdm(test_loader)):
with torch.no_grad():
A = A.cuda()
B = B.cuda()
Y = mask.cuda()
preds = net(A,B)
output = F.sigmoid(preds[1])
output[output >= 0.5] = 1
output[output < 0.5] = 0
pred = output.data.cpu().numpy().astype(int)
target = Y.cpu().numpy()
for i in range(output.shape[0]):
probs_array = (torch.squeeze(output[i])).data.cpu().numpy()
final_mask = probs_array * 255
final_mask = final_mask.astype(np.uint8)
final_savepath = save_path + filename[i] + '.png'
im = Image.fromarray(final_mask)
im.save(final_savepath)
Eva_test.add_batch(target, pred)
print('target.shape', target.shape)
print('pred.shape', pred.shape)
IoU = Eva_test.Intersection_over_Union()
Pre = Eva_test.Precision()
Recall = Eva_test.Recall()
F1 = Eva_test.F1()
OA=Eva_test.OA()
Kappa=Eva_test.Kappa()
# print('[Test] IoU: %.4f, Precision:%.4f, Recall: %.4f, F1: %.4f' % (IoU[1], Pre[1], Recall[1], F1[1]))
print('[Test] F1: %.4f, Precision:%.4f, Recall: %.4f, OA: %.4f, Kappa: %.4f,IoU: %.4f' % ( F1[1],Pre[1],Recall[1],OA[1],Kappa[1],IoU[1]))
# print('F1-Score: {:.2f}\nPrecision: {:.2f}\nRecall: {:.2f}\nOA: {:.2f}\nKappa: {:.2f}\nIoU: {:.2f}\n}'.format(F1[1] * 100, Pre[1] * 100, Recall[1] * 100, OA[1] * 100, Kappa[1] * 100, IoU[1] * 100))
print('F1-Score: Precision: Recall: OA: Kappa: IoU: ')
# print('{:.2f}\{:.2f}\{:.2f}\{:.2f}\{:.2f}\{:.2f}'.format(F1[1] * 100, Pre[1] * 100, Recall[1] * 100, OA[1] * 100, Kappa[1] * 100,IoU[1] * 100))
print('{:.2f} {:.2f} {:.2f} {:.2f} {:.2f} {:.2f}\n'.format(F1[1] * 100, Pre[1] * 100, Recall[1] * 100, OA[1] * 100, Kappa[1] * 100,IoU[1] * 100))
print('{:.2f} {:.2f} {:.2f} {:.2f} {:.2f} {:.2f}\n'.format(F1[0] * 100, Pre[0] * 100, Recall[0] * 100, OA[0] * 100, Kappa[0] * 100,IoU[0] * 100))
if __name__ == '__main__':
import argparse
# 'SemiModel_without_GCM_3_4_5'
# "SemiModel_without_GCM_1_2_3"
# "SemiModel_without_all_GCM"
# "SemiModel_without_Refine"
# 'SemiModel_without_agg_init'
# 'SemiModel_without_agg_final'
# 'Full'
Ablation_name = 'SemiModel_without_GCM_3_4_5'
parser = argparse.ArgumentParser()
parser.add_argument('--batchsize', type=int, default=16, help='training batch size')
parser.add_argument('--trainsize', type=int, default=256, help='training dataset size')
parser.add_argument('--gpu_id', type=str, default='2', help='train use gpu') #修改这里！！！
parser.add_argument('--data_name', type=str, default='LEVIR', #修改这里！！！
help='the test rgb images root')
parser.add_argument('--model_name', type=str, default='SemiModel_noema04', #修改这里！！！
help='the test rgb images root')
# parser.add_argument('--save_path', type=str, default='/data/chengxi.han/Sigma122/HANet_Model3-Semi/output/test_wokers/') # 半监督
# parser.add_argument('--save_path', type=str, default='./test_result/WHU/WHU-labeled-10-semiloss-1.1-2/')
# parser.add_argument('--save_path', type=str, default='./test_result/LEVIR/LEVIR-WHU-Student-30%-100Epo/')
# parser.add_argument('--save_path', type=str, default='./test_result/S2Looking/S2Looking-supervised-100-Teacher/')
# parser.add_argument('--save_path', type=str, default='./test_result-supervised/SYSU/SemiHANet-SYSU-Sepuervised-100%-Teacher/')
# parser.add_argument('--save_path', type=str, default='./test_result/LEVIR/SemiHANet-WHU-LEVIR-Teacher-30%-100Epo-F1/')
# parser.add_argument('--save_path', type=str,default='./test_result/LEVIR/SemiHANet-WHU-LEVIR-Student-30%-100Epo-F1/')
# parser.add_argument('--save_path', type=str,default='./test_result/LEVIRsup-WHUunsup/SemiHANet-LEVIRsup-WHUunsup-Teacher-30/')
# parser.add_argument('--save_path', type=str,default='./test_result/WHUsup-LEVIRunsup/SemiHANet-WHUsup-LEVIRunsup-Student-30/')
parser.add_argument('--save_path', type=str, default='./test_result/AblationStudy/LEVIR-5-without_GCM_A_B_C-Teacher/')
# 半监督影像保存路径
# parser.add_argument('--save_path', type=str, default='./test_result-supervised/') #全监督的影像保存路径
opt = parser.parse_args()
# set the device for training
if opt.gpu_id == '0':
os.environ["CUDA_VISIBLE_DEVICES"] = "0"
print('USE GPU 0')
elif opt.gpu_id == '1':
os.environ["CUDA_VISIBLE_DEVICES"] = "1"
print('USE GPU 1')
if opt.gpu_id == '2':
os.environ["CUDA_VISIBLE_DEVICES"] = "2"
print('USE GPU 2')
if opt.gpu_id == '3':
os.environ["CUDA_VISIBLE_DEVICES"] = "3"
print('USE GPU 3')
if opt.data_name == 'LEVIR':
opt.test_root = '/data/chengxi.han/data/LEVIR CD Dataset256/test/'
# opt.test_root = '/data/chengxi.han/data/LEVIR-CD-Dataset256/test256-sub/'
# opt.test_root = '/data/chengxi.han/data/LEVIR CD Dataset256/test256-sub-331/'
# opt.test_root = '/data/chengxi.han/data/LEVIR-CD-Dataset256/test256-sub-21/'
# opt.test_root = '/data/chengxi.han/data/LEVIR-CD-Dataset256/test256-sub-1419/'
elif opt.data_name == 'WHU':
# opt.test_root = '/data/chengxi.han/data/Building change detection dataset256/test/'
opt.test_root = '/data/chengxi.han/data/WHU-CD-256-Semi/test/'
# opt.test_root = '/data/chengxi.han/data/Building change detection dataset256/test-sub/'
# opt.test_root = '/data/chengxi.han/data/Building change detection dataset256/test-sub-61-3/'
elif opt.data_name == 'CDD':
opt.test_root = '/data/chengxi.han/data/CDD_ChangeDetectionDataset/Real/subset/test/'
# opt.test_root = '/data/chengxi.han/data/Building change detection dataset256/test-sub-CDD402/'
elif opt.data_name == 'DSIFN':
opt.test_root = '/data/chengxi.han/data/DSIFN256/test/'
elif opt.data_name == 'SYSU':
opt.test_root = '/data/chengxi.han/data/SYSU-CD/test/'
# opt.test_root = '/data/chengxi.han/data/SYSU-CD/test-sub/'
elif opt.data_name == 'S2Looking':
opt.test_root = '/data/chengxi.han/data/S2Looking256/test/'
# opt.test_root = '/data/chengxi.han/data/S2Looking256/test-sub/'
elif opt.data_name == 'GoogleGZ':
opt.test_root = '/data/chengxi.han/data/Google_GZ_CD256/test/'
elif opt.data_name == 'LEVIRsup-WHUunsup':
opt.test_root = '/data/chengxi.han/data/WHU-LEVIR-CD-256-Semi/test/'
opt.save_path = opt.save_path + opt.data_name + '/' + opt.model_name + '/'
test_loader = data_loader.get_test_loader(opt.test_root, opt.batchsize, opt.trainsize, num_workers=2, shuffle=False, pin_memory=True)
Eva_test = Evaluator(num_class=2)
if opt.model_name == 'HANet_v2':
model = HANet_v2().cuda()
elif opt.model_name == 'HANet_v3':
model = HANet_v3().cuda()
elif opt.model_name == 'HANet_v4':
model = HANet_v4().cuda()
elif opt.model_name == 'HANet_v5':
model = HANet_v5().cuda()
elif opt.model_name == 'HANet_v6':
model = HANet_v6().cuda()
elif opt.model_name == 'HANet_v6_Ablation':
model = HANet_v6_Ablation().cuda()
elif opt.model_name == 'HANModel3':
model = HANModel3().cuda()
elif opt.model_name == 'SemiModel_noema04':
model = SemiModel().cuda()
if Ablation_name == 'SemiModel_without_GCM_3_4_5':
model = Ablation_model.SemiModel_without_GCM_3_4_5().cuda()
ema_model = Ablation_model.SemiModel_without_GCM_3_4_5().cuda()
elif Ablation_name == "SemiModel_without_GCM_1_2_3":
model = Ablation_model.SemiModel_without_GCM_1_2_3().cuda()
ema_model = Ablation_model.SemiModel_without_GCM_1_2_3().cuda()
elif Ablation_name == "SemiModel_without_all_GCM":
model = Ablation_model.SemiModel_without_all_GCM().cuda()
ema_model = Ablation_model.SemiModel_without_all_GCM().cuda()
elif Ablation_name == "SemiModel_without_Refine":
model = Ablation_model.SemiModel_without_Refine().cuda()
ema_model = Ablation_model.SemiModel_without_Refine().cuda()
elif Ablation_name == 'SemiModel_without_agg_init':
model = Ablation_model.SemiModel_without_agg_init().cuda()
ema_model = Ablation_model.SemiModel_without_agg_init().cuda()
elif Ablation_name == 'SemiModel_without_agg_final':
model = Ablation_model.SemiModel_without_agg_final().cuda()
ema_model = Ablation_model.SemiModel_without_agg_final().cuda()
elif Ablation_name == 'Full':
model = SemiModel().cuda()
ema_model = SemiModel().cuda()
# opt.load = './output/' + opt.data_name + '/' + opt.model_name + '_best_iou.pth'
# opt.load = './output/' + opt.data_name + '/' + opt.model_name + '_best_student_iou.pth' #半监督
# opt.load = './output/' + opt.data_name + '-5%/' + opt.model_name + '_best_teacher_iou.pth' #半监督
# opt.load = './output-supervised/' + opt.data_name + '/' + opt.model_name + '_best_student_iou.pth' # #全监督的
# opt.load = './output-supervised/' + opt.data_name + '/' + opt.model_name + '_best_teacher_iou.pth' # #全监督的
# iou 83.84 10% GoogleGZ
# opt.load = './output/test_wokers/load1' + opt.data_name + '/' + opt.model_name + '_best_teacher_iou.pth' #半监督
# parser.add_argument('--save_path', type=str,default='./output/test_wokers/load1/') #半监督的路径
#-------测试老师的模型teacher-------
# save_path = './output/S2Looking-5/'
# save_path = './output/SYSU-Sepuervised-100%/'
# save_path = './output/test_wokers/load1/WHU-30-100Epo/'
# save_path = './output-supervised/SYSU-30/' # 全监督的
# save_path = './output/LEVIRsup-WHUunsup-30/'
# save_path = './output/WHUsup-LEVIRunsup-30/'
save_path = './output/LEVIR-5-without_GCM_A_B_C/'
# save_path = save_path + '/' + opt.model_name
save_path = save_path + opt.data_name + '/' + opt.model_name
opt.load = save_path + '_train1_' + '_best_teacher_iou.pth'
if opt.load is not None:
model.load_state_dict(torch.load(opt.load))
print('load model from ', opt.load)
# print('路径是：WHU-labeled-5-semiloss-0.8')
#-------测试学生的模型student-------
# # save_path = './output/S2Looking-5/'
# # save_path = './output/SYSU-Sepuervised-100%/'
# # save_path = './output/test_wokers/load1/' #半监督的
# # save_path = './output-supervised/LEVIR-5' #全监督的
# # save_path = './output/test_wokers/load1/WHU-30-100Epo'
# # save_path = './output/LEVIRsup-WHUunsup-30/'
# # save_path = './output/WHUsup-LEVIRunsup-30/'
# save_path = './output/S2Looking-supervised-100/'
#
# # save_path = save_path + opt.model_name
# save_path = save_path + opt.data_name + '/' + opt.model_name
# # opt.load = save_path + opt.data_name + '/' + opt.model_name+ '_train1_' + '_best_student_iou.pth'
#
# # save_path = save_path + opt.model_name
# opt.load = save_path + '_train1_'+ '_best_student_iou.pth'
# # opt.load ='./output/LEVIR-5%/SemiModel_noema04_best_teacher_iou.pth'
# if opt.load is not None:
# print('load model from ', opt.load)
# checkpoint_stud = torch.load(opt.load)
# model.load_state_dict(checkpoint_stud['best_student_net '])
#
save_path = opt.save_path
if not os.path.exists(save_path):
os.makedirs(save_path)
test(test_loader, Eva_test, opt.save_path, model)
end=time.time()
print('程序测试test的时间为:',end-start)