-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathvisualizer.py
More file actions
224 lines (198 loc) · 9.71 KB
/
Copy pathvisualizer.py
File metadata and controls
224 lines (198 loc) · 9.71 KB
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
import logging
import numpy as np
import os
import sys
from subprocess import Popen, PIPE
from code_util import util
from code_util.data import read_save
from code_util.util import tensor2im, get_module_by_name
class Visualizer():
"""This class includes several functions that can display/save images and print/save logging information.
It uses a Python library 'visdom' for display, and a Python library 'dominate' (wrapped in 'HTML') for creating HTML files with images.
"""
def __init__(self, config):
"""Initialize the Visualizer class
Parameters:
Step 1: Cache the training/test options
Step 3: create an HTML object for saveing HTML filters
Step 4: create a logging file to store training losses
对train和test 有不同的处理
"""
self.config = config # cache the option
self.name = config["name"]
self.work_dir = config["work_dir"]
self.init_log()
# html
if self.config["record"].get("html",{}).get("use_html") == True:
self.use_html = True
self.init_html()
# tensorboardX
if self.config["record"].get("tensorboard", {}).get("use_tensorboard") == True:
self.use_tensorboard = True
self.init_tensorboard()
def init_log(self):
format = logging.Formatter('%(asctime)s - %(levelname)s - %(message)s')
if self.config["phase"] == "train":
self.logger_train = logging.getLogger("logger_train")
self.log_train_file = os.path.join(self.work_dir, 'train_log.txt')
self.handler_train = logging.FileHandler(self.log_train_file)
self.handler_train.setFormatter(format)
self.logger_train.setLevel(logging.INFO)
self.logger_train.addHandler(self.handler_train)
print(f"Training log initialized: {self.log_train_file}")
self.logger_train.info(f"Training started")
self.logger_val = logging.getLogger("logger_val")
self.log_val_file = os.path.join(self.work_dir, 'val_log.txt')
self.handler_val = logging.FileHandler(self.log_val_file)
self.handler_val.setFormatter(format)
self.logger_val.setLevel(logging.INFO)
self.logger_val.addHandler(self.handler_val)
print(f"Training log initialized: {self.log_val_file}")
self.logger_val.info(f"Validation started")
elif self.config["phase"] == "test":
self.logger_test = logging.getLogger("logger_test")
self.log_test_file = os.path.join(self.work_dir, 'test_log.txt')
self.handler_test = logging.FileHandler(self.log_test_file)
self.handler_test.setFormatter(format)
self.logger_test.setLevel(logging.INFO)
self.logger_test.addHandler(self.handler_test)
print(f"Test log initialized: {self.log_test_file}")
self.logger_test.info(f"Test started")
else:
raise ValueError("Invalid phase. Expected 'train' or 'test', got: %s" % self.config["phase"])
def init_html(self):
from . import html
self.web_dir = os.path.join(self.work_dir,'web')
print("create web directory:", self.web_dir)
os.makedirs(self.web_dir,exist_ok=True)
self.win_size = self.config["record"]["html"]["display_size"]
if self.config["phase"] == "train":
img_train_dir = os.path.join(self.web_dir, 'train')
os.makedirs(img_train_dir,exist_ok=True)
title = '%s | train' % self.work_dir
self.webpage_train = html.HTML(self.web_dir, title, filename = "train", refresh=0)
img_val_dir = os.path.join(self.web_dir, 'val')
os.makedirs(img_val_dir,exist_ok=True)
title = '%s | val' % self.work_dir
self.webpage_val = html.HTML(self.web_dir, title, filename = "val", refresh=0)
self.img_dir = [img_train_dir, img_val_dir]
else:
img_test_dir = os.path.join(self.web_dir, 'test')
os.makedirs(img_test_dir,exist_ok=True)
title = 'Experiment = %s, Epoch = %s' % (self.name, self.config["result"]["test_epoch"])
self.webpage_test = html.HTML(self.web_dir, title, filename = "test", refresh=0)
self.img_dir = [img_test_dir]
def init_tensorboard(self):
"""Initialize TensorBoardX writer."""
from tensorboardX import SummaryWriter # 添加 tensorboardX 的导入
tensorboard_dir = os.path.join(self.work_dir, 'tensorboard_logs')
os.makedirs(tensorboard_dir, exist_ok=True)
self.writer = SummaryWriter(log_dir=tensorboard_dir)
print(f"TensorBoard logs initialized: {tensorboard_dir}")
# # 启动tensorboard 会和tqdm的显示冲突 暂时注释掉
# from tensorboard import program
# tb = program.TensorBoard()
# tb.configure(argv=[None, '--logdir', tensorboard_dir])
# url = tb.launch()
# print(f"TensorBoard 启动成功，访问链接: {url}")
def init_CAM(self):
import cv2
from code_util.cam.grad_cam import GradCAM, show_cam_on_image
def display_on_html(self, visuals, img_paths, phase = 'train', epoch = 0, iter = 0, ):
"""
save current results to an HTML file.
"""
file_name = util.get_file_name(img_paths[0])
if phase == "train":
time_info = f"epoch {epoch} iter {iter}, {file_name}"
save_img_path_clip = '%s_%s_%s' % (epoch, iter, file_name)
img_dir = self.img_dir[0]
webpage = self.webpage_train
elif phase == "val":
time_info = f"epoch {epoch}, {file_name}"
save_img_path_clip = '%s_%s' % (epoch, file_name)
img_dir = self.img_dir[1]
webpage = self.webpage_val
else: # test
test_epoch = self.config["result"]["test_epoch"]
time_info = f"test epoch {test_epoch}, {file_name}"
save_img_path_clip = file_name
img_dir = self.img_dir[0]
webpage = self.webpage_test
# save images to the disk
for label, image in visuals.items():
image_numpy = util.tensor2im(image, return_first=True)
save_img_path = os.path.join(img_dir, save_img_path_clip + '_' + label + '.png')
read_save.save_image_4_show(image_numpy, save_img_path)
# update website
webpage.add_header(time_info)
ims, txts, links = [], [], []
for label, image_numpy in visuals.items():
save_image_name = save_img_path_clip + '_' + label + '.png'
ims.append(save_image_name)
txts.append(label)
links.append(save_image_name)
webpage.add_images(ims, txts, links, width=self.win_size)
webpage.save()
def display_on_tensorboard(self, visuals, step, phase='train'):
"""
Log images to TensorBoard.
Parameters:
visuals (dict) -- dictionary of images to log
epoch (int) -- current epoch
phase (str) -- 'train' or 'val'
iter (int) -- current iteration (used for 'train' phase)
"""
for label, image in visuals.items():
image_numpy = util.tensor2im(image) # Convert tensor to numpy image(0~255)
if phase == 'train':
self.writer.add_image(f"{phase}/{label}", image_numpy, global_step=step, dataformats='HWC')
elif phase == 'val':
self.writer.add_image(f"{phase}/{label}", image_numpy, global_step=step, dataformats='HWC')
def plot_scalars_on_tensorboard(self, scalars, epoch, phase='train'):
"""
Log scalars to TensorBoard.
Parameters:
scalars (dict) -- dictionary of scalars to log
epoch (int) -- current epoch
phase (str) -- 'train' or 'val'
"""
for scalar_name, scalar_value in scalars.items():
self.writer.add_scalar(f"{phase}/{scalar_name}", scalar_value, global_step=epoch)
def close_tensorboard(self):
"""Close the TensorBoard writer."""
if hasattr(self, 'writer'):
self.writer.close()
# losses: same format as |losses| of plot_current_losses
def record_log(self, info, phase):
if phase == "train":
self.logger_train.info(info)
elif phase == "val":
self.logger_val.info(info)
elif phase == "test":
self.logger_test.info(info)
else:
raise ValueError("Invalid phase. Expected 'train', 'val' or 'test', got: %s" % phase)
def draw_CAM(self, model, config, epoch=None, img_paths=None):
"""Draw and save CAM images."""
if img_paths is None:
img_paths = model.get_image_paths()
target_layer = get_module_by_name(model.netG, config["record"]["CAM"]["CAM_layer"])
grad_cam = GradCAM(model.netG, target_layers=[target_layer], move2cuda=False)
grayscale_cam = grad_cam(input_tensor=model.real_A, target=model.real_B)
real_B = tensor2im(model.real_B)
# Save CAM images
save_path = os.path.join(config["work_dir"], "CAM")
os.makedirs(save_path, exist_ok=True)
A_path = img_paths["A_path"]
if isinstance(A_path, list):
A_path = A_path[0]
save_name = A_path.split("/")[-1].split(".")[0]
if epoch is not None:
save_name = save_name + "_epoch_" + str(epoch)
save_name = save_name + "_CAM.jpg"
save_path = os.path.join(save_path, save_name)
# Save as grayscale image
cam_real = show_cam_on_image(real_B / 255, grayscale_cam[0, :, :], use_rgb=True)
cv2.imwrite(save_path, cam_real)
self.logger_train.info(f"CAM image saved at: {save_path}")