forked from Jack-GVDL/FastRCNN
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathTrain_Alexnet.py
199 lines (159 loc) · 6.61 KB
/
Train_Alexnet.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
import time
from datetime import datetime
from typing import *
import torch
import torch.optim as optim
from Lib import *
# ----- check environment -----
# env_device = torch.cuda.current_device()
env_device = "cuda:0" if torch.cuda.is_available() else "cpu"
env_device = torch.device(env_device)
print(f"----- Environment -----")
print(f"CUDA available: {torch.cuda.is_available()}")
print(f"Device name: {torch.cuda.get_device_name(env_device)}")
print(f"Device memory: {torch.cuda.get_device_properties(env_device).total_memory}")
# ----- model -----
info = ModelInfo()
info.epoch = 6
info.batch_size = 1
info.learning_rate = 5e-4
info.train_parameter["Momentum"] = 0.9
info.device_train = env_device
info.device_test = env_device
info.model = FastRCNN_Alexnet()
info.optimizer = optim.SGD(info.model.parameters(), lr=info.learning_rate, momentum=info.train_parameter["Momentum"])
# info.optimizer = optim.Adam(pack.net.parameters(), lr=1e-4)
# info.scheduler = optim.lr_scheduler.StepLR(pack.optimizer, step_size=30, gamma=0.1)
info.scheduler = None
print(f"----- Model -----")
print(info.model)
# ----- dataset -----
# data file
file_config_train = "./Data/train/Data_20201225212039_0.75_1.0_0.1_0.75_320_320.json"
file_config_val = "./Data/val/Data_20201226131228_0.75_1.0_0.1_0.75_32_32.json"
file_config_test = "./Data/test/Data_20201223123013_0.5_1.0_0.1_0.5_32_32.json"
data_path_train = "./Data/train"
data_path_val = "./Data/val"
data_path_test = "./Data/test"
# config
config_train = Config_Processed()
config_train.file = file_config_train
config_train.load()
config_validate = Config_Processed()
config_validate.file = file_config_val
config_validate.load()
config_test = Config_Processed()
config_test.file = file_config_test
config_test.load()
# dataset
dataset_train = Dataset_Processed(config_train, data_path=data_path_train)
dataset_validate = Dataset_Processed(config_validate, data_path=data_path_val)
dataset_test = Dataset_Processed(config_test, data_path=data_path_test)
dataset_train.size_positive = 16
dataset_train.size_negative = 16
dataset_validate.size_positive = 16
dataset_validate.size_negative = 16
# dataset list
dataset_list: Dict = {
"Train": dataset_train,
"Val": dataset_validate,
"Test": dataset_test
}
# save to info
info.train_parameter["DataTrain"] = file_config_train
info.train_parameter["DataVal"] = file_config_val
info.train_parameter["DataTest"] = file_config_test
print(f"----- Dataset -----")
print(info.train_parameter["DataTrain"])
print(info.train_parameter["DataVal"])
print(info.train_parameter["DataTest"])
# ----- process -----
now = datetime.now()
current_time = now.strftime("%Y%m%d%H%M%S")
info.save_path = "./Result"
info.save_folder = f"Result_{current_time}"
# create process
process_hook_result = TrainProcess_Hook()
process_counter = TrainProcess_Counter()
process_folder = TrainProcess_Folder()
process_result_data = TrainProcess_ResultData()
process_python_file = TrainProcess_PythonFile()
process_dict_save = TrainProcess_DictSave()
process_result_graph = TrainProcess_ResultGraph()
process_result_record_1 = TrainProcess_ResultRecord()
process_result_record_2 = TrainProcess_ResultRecord()
# config stage
process_folder.addStage( ModelInfo.Stage.TRAIN_START)
process_python_file.addStage( ModelInfo.Stage.TRAIN_START)
process_counter.addStage( ModelInfo.Stage.ITERATION_VAL_END)
process_result_data.addStage( ModelInfo.Stage.ITERATION_VAL_END)
process_result_record_1.addStage( ModelInfo.Stage.ITERATION_VAL_END)
process_dict_save.addStage( ModelInfo.Stage.TRAIN_END)
process_hook_result.addStage( ModelInfo.Stage.TRAIN_END)
process_result_graph.addStage( ModelInfo.Stage.TRAIN_END)
process_result_record_2.addStage( ModelInfo.Stage.TRAIN_END)
process_result_data.accuracy_index = 5 # 5 is the confusion matrix of val_total
process_python_file.addPythonFile(info.model, "FastRCNN_Alexnet.py")
process_dict_save.addDictData(info, "ModelInfo.json")
process_result_graph.addAccuracy([3, 4, 5], ["Label", "Box", "Total"], "Accuracy.png")
process_result_graph.addLoss([2, 5], ["Train", "Val"], "Loss.png")
process_counter.addProcess(process_result_record_1, 2)
process_result_record_1.result = process_result_data
process_result_record_2.result = process_result_data
# TODO: may move to other place
# Linker Function
def Linker_Hook_execute(stage: int, info_: ModelInfo, data: Dict) -> None:
process_result_graph.addConfusionMatrix(
(process_result_data.best_epoch, 3),
( ["Predict-NIL", "Predict-MOD", "Predict-SEV"],
["Ground-NIL", "Ground-MOD", "Ground-SEV"]),
"ConfusionMatrix_Class.png")
process_result_graph.addConfusionMatrix(
(process_result_data.best_epoch, 4),
( ["Predict-MOD-T", "Predict-MOD-F", "Predict-SEV-T", "Predict-SEV-F"],
["Ground-MOD-T", "Ground-MOD-F", "Ground-SEV-T", "Ground-SEV-F"]),
"ConfusionMatrix_IOU.png")
process_result_graph.addConfusionMatrix(
(process_result_data.best_epoch, 5),
( ["Predict-NIL", "Predict-MOD", "Predict-SEV", "Predict-IOU-F"],
["Ground-NIL", "Ground-MOD", "Ground-SEV", "Ground-IOU-F"]),
"ConfusionMatrix_Total.png")
# when process_resultGraph is called
# it should generate the confusion matrix of best epoch
process_hook_result.func_execute = Linker_Hook_execute
process_python_file.is_print = True
process_dict_save.is_print = True
process_result_data.is_print = True
process_result_record_1.is_print = True
process_result_record_2.is_print = True
process_result_graph.is_print = True
process_result_data.is_log = True
process_result_record_1.is_log = True
process_result_record_2.is_log = True
process_result_graph.is_log = True
info.process_control.addProcess(process_folder)
info.process_control.addProcess(process_python_file)
info.process_control.addProcess(process_result_data)
info.process_control.addProcess(process_counter)
info.process_control.addProcess(process_dict_save)
info.process_control.addProcess(process_hook_result)
info.process_control.addProcess(process_result_graph)
info.process_control.addProcess(process_result_record_2)
# probe
probe = TrainProcessProbe()
probe.process_control = info.process_control
probe.probe()
# TODO: test
print(probe.getLogContent(0, None))
breakpoint()
# ----- train -----
print(f"----- Train -----")
# config model
info.model.setGradient(FastRCNN_Alexnet.Layer.INPUT_CONV, False)
info.model.setGradient(FastRCNN_Alexnet.Layer.ALEXNET, False)
info.model.setGradient(FastRCNN_Alexnet.Layer.POOL, False)
info.model.setGradient(FastRCNN_Alexnet.Layer.FEATURE, True)
info.model.setGradient(FastRCNN_Alexnet.Layer.SOFTMAX, True)
info.model.setGradient(FastRCNN_Alexnet.Layer.BOX, True)
info.model.is_detach = True
train(dataset_list, info)