-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathtrain.py
222 lines (178 loc) · 9.69 KB
/
train.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
import random
import numpy as np
import torch
from torch.utils.tensorboard import SummaryWriter
# from models.STG_NF.Unet import Unet
from models.STG_NF.model_pose import STG_NF
from models.pretrain_NF import Trainer
from utils.data_utils import trans_list
from utils.optim_init import init_optimizer, init_scheduler
from args import create_exp_dirs
from args import init_parser, init_sub_args
from dataset import get_dataset_and_loader
from utils.train_utils import dump_args, init_model_params
from utils.scoring_utils import score_dataset
from utils.train_utils import calc_num_of_params
from models.unamsk_encoder.RGBEnc import RGB_Encoder
from models.train_RGBEnc import RGBEnc_Trainer
import matplotlib.pyplot as plt
import time
from dataset import PoseSegDataset
import os
# os.environ["CUDA_VISIBLE_DEVICES"] = '1'
# os.environ['CUDA_LAUNCH_BLOCKING'] = '1'
def plot_and_save(x,y,auc,log_path,save_path):
plt.plot(x,y)
now_time = time.time()
with open(log_path, "a") as f:
f.write( str(auc * 100) + "%\n")
plt.savefig(save_path+"%s"% now_time+".jpg", dpi=300)
def main():
parser = init_parser()
args = parser.parse_args()
root = os.getcwd()
if args.dataset == "ShanghaiTech":
feat_path_root = os.path.join(root, r"../STG-NF-main/data/ShanghaiTech/feat")
if args.tea_pretrained is None:
tea_pretrained = os.path.join(root,"checkpoints/ShanghaiTech_85_9.tar")
else:
tea_pretrained = os.path.join(args.tea_pretrained)
else:
feat_path_root = os.path.join(root, r"../STG-NF-main/data/UBnormal/features/ubnormal_feature_i3d")
if args.tea_pretrained is None:
tea_pretrained = os.path.join(root,"checkpoints/UBnormal_unsupervised_71_8.tar")
else :
tea_pretrained = os.path.join(args.tea_pretrained)
if args.seed == 999: # Record and init seed
args.seed = torch.initial_seed()
np.random.seed(0)
else:
random.seed(args.seed)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = True
torch.manual_seed(args.seed)
np.random.seed(0)
args, model_args = init_sub_args(args)
args.ckpt_dir = create_exp_dirs(args.exp_dir, dirmap=args.dataset)
pretrained = vars(args).get('checkpoint', None)
dataset, loader = get_dataset_and_loader(args, trans_list=trans_list, feat_path_root = feat_path_root,only_test=(pretrained is not None),only_train = False ) ## 这里产生数据集 #
# dataset :[(3,24,18),trans_index,segs_score_np(24,),labels(1,)]
'''
1. Load data from JSON files.
2. Process each JSON file on a per-character basis in a loop: gen_dataset
- First, obtain all frames related to a character.
- Next, pack a certain number of consecutive frames into a group based on seg_len (24) and overlap.
3. For each call to the dataset's __getitem__, data is organized in segments (segs): [(3, 24, 18), trans_index, segs_score_np(24,), labels(1,)].
4. Both the training and testing processes operate on a per-segment basis.
5. When calculating AUC, classify each segment to the corresponding frame in scene_clip_.
'''
model_args = init_model_params(args, dataset)
print(model_args)
tea_model = STG_NF(**model_args)
num_of_params = calc_num_of_params(tea_model)
trainer = Trainer(args, tea_model, loader['train'], loader['test'],
optimizer_f=init_optimizer(args.model_optimizer, lr=args.model_lr),
scheduler_f=init_scheduler(args.model_sched, lr=args.model_lr, epochs=args.epochs)) ## 训练器
if args.stage == 1:
print("Stage One : Train pretrained Model")
if pretrained:
trainer.load_checkpoint(pretrained)
else:
writer = SummaryWriter()
trainer.train(log_writer=writer)
normality_scores = trainer.test()
auc, scores ,cal_abnormal_nums= score_dataset(normality_scores, dataset["test"].metadata, args=args)
# Logging and recording results
print("\n-------------------------------------------------------")
print("\033[92m Done with {}% AuC for {} samples\033[0m".format(auc * 100, scores.shape[0]))
with open("log/log.txt", "a") as f:
f.write(str(args.epochs)+"\t" + str(args.K) + "\t" + str(args.L) + "\t"+ str(args.seg_len) + "\t"+ str(args.seg_stride) + "\t" + str(args.model_lr) + "\t"+ str(args.model_lr_decay) + "\t"+ str(auc * 100) + "%\n")
###################### Stage 3 : Train Feat_Student Model ######################
if args.stage == 2 :
tea_pretrained = tea_pretrained
print("Stage 3 : Train Feat_Student Model")
checkpoint = torch.load(tea_pretrained)
tea_model.load_state_dict(checkpoint['state_dict'], strict=False)
tea_model.set_actnorm_init()
model_args = init_model_params(args, dataset)
feat_Encoder_model = RGB_Encoder(1024)
feat_Encoder_trainer = RGBEnc_Trainer(args, feat_Encoder_model, tea_model,loader['train'], loader['test'],
optimizer_f=init_optimizer(args.model_optimizer, lr=args.model_lr),
scheduler_f=init_scheduler(args.model_sched, lr=args.model_lr, epochs=args.epochs)) ## 训练器
if pretrained: ## 加载预训练模型 /ssd/agqing/STG-NF-main/data/exp_dir/ShanghaiTech/Apr20_1139/Apr20_1143__checkpoint.pth.tar
feat_stu_trainer.feat_load_checkpoint(pretrained)
normality_scores = feat_Encoder_trainer.feat_test() ## 训练器的测试过程
auc, scores, cal_abnormal_nums = score_dataset(normality_scores, dataset["test"].metadata, args=args)
print("\n-------------------------------------------------------")
print("\033[92m Done with {}% AuC for {} samples\033[0m".format(auc * 100, scores.shape[0]))
print("-------------------------------------------------------\n\n")
plot_and_save(cal_abnormal_nums[:, 0], cal_abnormal_nums[:, 2], auc, log_path="log/log2.txt",
save_path="log/test_stage2_abnormal_rate_%s.png")
else:
writer = SummaryWriter()
pretrained = feat_Encoder_trainer.feat_train(log_writer=writer) ## 开始训练
return pretrained #
def test(pretrained,batch_size):
parser = init_parser()
args = parser.parse_args()
args.batch_size = batch_size
args.checkpoint = "checkpoints/feat_checkpoint/" +pretrained
root = os.getcwd()
if args.dataset == "ShanghaiTech":
feat_path_root = os.path.join(root, r"data/ShanghaiTech/feat")
if args.tea_pretrained is None:
tea_pretrained = os.path.join(root, "checkpoints/pretrained_ShanghaiTech.tar")
else:
tea_pretrained = os.path.join(args.tea_pretrained)
else:
feat_path_root = os.path.join(root, r"data/UBnormal/features/ubnormal_feature_i3d")
if args.tea_pretrained is None:
tea_pretrained = os.path.join(root, "checkpoints/pretrained_UBnormal.tar")
else:
tea_pretrained = os.path.join(args.tea_pretrained)
if args.seed == 999: # Record and init seed
args.seed = torch.initial_seed()
np.random.seed(0)
else:
random.seed(args.seed)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = True
torch.manual_seed(args.seed)
np.random.seed(0)
args, model_args = init_sub_args(args)
args.ckpt_dir = create_exp_dirs(args.exp_dir, dirmap=args.dataset)
pretrained = vars(args).get('checkpoint', None)
dataset, loader = get_dataset_and_loader(args, trans_list=trans_list, feat_path_root=feat_path_root,
only_test=(pretrained is not None), only_train=False) ## 这里产生数据集 #
model_args = init_model_params(args, dataset)
print(model_args)
tea_model = STG_NF(**model_args) ## 模型加载
num_of_params = calc_num_of_params(tea_model)
trainer = Trainer(args, tea_model, loader['train'], loader['test'],
optimizer_f=init_optimizer(args.model_optimizer, lr=args.model_lr),
scheduler_f=init_scheduler(args.model_sched, lr=args.model_lr, epochs=args.epochs)) ## 训练器
###################### Stage 2 : Train Feat_Student Model ######################
if args.stage == 2:
tea_pretrained = tea_pretrained
print("Stage 2 : Train Feat_Student Model")
checkpoint = torch.load(tea_pretrained)
tea_model.load_state_dict(checkpoint['state_dict'], strict=False)
tea_model.set_actnorm_init()
# optimizer.load_state_dict(checkpoint['optimizer'])
print(pretrained)
model_args = init_model_params(args, dataset)
feat_Encoder_model = RGB_Encoder(1024)
feat_Encoder_trainer = RGBEnc_Trainer(args, feat_Encoder_model, tea_model, loader['train'], loader['test'],
optimizer_f=init_optimizer(args.model_optimizer, lr=args.model_lr),
scheduler_f=init_scheduler(args.model_sched, lr=args.model_lr,
epochs=args.epochs))
feat_stu_trainer.feat_load_checkpoint(pretrained)
normality_scores = feat_stu_trainer.feat_test()
auc, scores ,cal_abnormal_nums= score_dataset(normality_scores, dataset["test"].metadata, args=args)
print("\n-------------------------------------------------------")
print("\033[92m Done with {}% AuC for {} samples\033[0m".format(auc * 100, scores.shape[0]))
print("-------------------------------------------------------\n\n")
if __name__ == '__main__':
pretrained = main()
print(pretrained)
test(pretrained, 32)