-
Notifications
You must be signed in to change notification settings - Fork 2
/
evaluate.py
360 lines (313 loc) · 13 KB
/
evaluate.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
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
import argparse
import os
import sys
import time
import yaml
import warnings
warnings.simplefilter("ignore", UserWarning)
import numpy as np
import torch
import torch.backends.cudnn as cudnn
import torch.nn as nn
import torch.nn.parallel
import torch.nn.functional as F
import torchvision.models as models
import matplotlib.pyplot as plt
from skimage import io, transform
from hair_data import GeneralDataset, get_helen_test_data, gen_transform_data_loader
from HairNet import DFN
from component.metrics import Acc_score
from tool_func import *
global args, device, save_dir
parser = argparse.ArgumentParser(
description='Pytorch Hair Segmentation Evaluate')
parser.add_argument(
'evaluate_name', type=str, help='evaluate name | that is save dir')
parser.add_argument(
'--model_name', required=True, default='', type=str, metavar='model name')
parser.add_argument('--batch_size', required=True, type=int, help='batch_size')
parser.add_argument(
'--save',
type=str2bool,
nargs='?',
default=False,
help='save or visualize')
parser.add_argument(
'--original',
type=str2bool,
nargs='?',
default=False,
help='evaluate on the original image')
parser.add_argument('--gpu_ids', type=int, nargs='*')
parser.add_argument(
'--data_settings', default='aug_512_0.6_multi_person', type=str)
parser.add_argument(
'--tform_back',
type=str2bool,
nargs='?',
default=True,
help='evaluate on transform back')
args = parser.parse_args()
device = None
save_dir = None
options = None
def main():
global device, save_dir, options
# use the gpu or cpu as specificed
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
device_ids = None
if args.gpu_ids is None:
if torch.cuda.is_available():
device_ids = list(range(torch.cuda.device_count()))
else:
device_ids = args.gpu_ids
device = torch.device("cuda:{}".format(device_ids[0]))
# set save dir
ROOT_DIR = os.path.dirname(os.path.abspath(__file__))
save_dir = os.path.join(ROOT_DIR, 'evaluate_' + args.evaluate_name)
os.makedirs(save_dir, exist_ok=True)
# set options path
option_path = os.path.join('logs', args.model_name,
args.model_name + '.yaml')
if not os.path.exists(option_path):
print('options path {} is not exists.'.format(option_path))
sys.exit(1)
options = yaml.load(open(option_path))
# check model path
model_path = os.path.join('logs', args.model_name, 'checkpoint.pth')
if not os.path.exists(model_path):
print('model path {} is not exists.'.format(model_path))
sys.exit(1)
# build the model
add_fc = options.get('add_fc', False)
self_attention = options.get('self_attention', False)
attention_plus = options.get('channel_attention', False)
in_channels = 3
if options.get('position_map', False):
in_channels = 5
elif options.get('center_map', False):
in_channels = 5
elif options.get('with_gaussian', False):
in_channels = 4
model = DFN(
in_channels=in_channels,
add_fc=add_fc,
self_attention=self_attention,
attention_plus=attention_plus,
back_bone=options['arch'])
model = nn.DataParallel(model, device_ids=device_ids)
model.to(device)
# loading checkpoint
print("=> loading checkpoint '{}'".format(model_path))
checkpoint = torch.load(model_path, map_location='cpu')
model.load_state_dict(checkpoint['state_dict'])
print("=> loaded checkpoint '{}' (epoch {})".format(
model_path, checkpoint['epoch']))
if options.get('position_map', False) or args.original:
test_ds = gen_transform_data_loader(
options,
mode='test',
batch_size=1,
shuffle=False,
dataloader=False,
use_original=args.original)
evaluate_general_dataset(model, test_ds)
else:
test_ds = get_helen_test_data(
['hair'], aug_setting_name=args.data_settings)
evaluate_raw_dataset(model, test_ds)
# ------ begin evaluate
def evaluate_raw_dataset(model, dataset):
global options
batch_time = AverageMeter()
acc_hist_all = Acc_score(['hair'])
acc_hist_single = Acc_score(['hair'])
mean = [0.485, 0.456, 0.406]
std = [0.229, 0.224, 0.225]
x_pos_map, y_pos_map, gaussian_map = None, None, None
channel_size = 3
if options.get('center_map', False):
x_pos_map, y_pos_map = GeneralDataset.get_xy_map(
dataset.load_image(0).shape[0])
x_pos_map = np.expand_dims(x_pos_map, -1)
y_pos_map = np.expand_dims(y_pos_map, -1)
channel_size = 5
elif options.get('with_gaussian', False):
gaussian_map = GeneralDataset.get_gaussian_map(
dataset.load_image(0).shape[0])
gaussian_map = np.expand_dims(gaussian_map, -1)
channel_size = 4
# switch to evaluate mode
model.eval()
with torch.no_grad():
end = time.time()
batch_index = 0
batch = None
labels = None
image_ids = []
image_names = []
data_len = len(dataset.image_ids)
for idx, image_id in enumerate(dataset.image_ids):
# ------ start iteration
image = dataset.load_image(image_id)
if batch_index == 0:
batch = np.zeros((args.batch_size, image.shape[0],
image.shape[1], channel_size))
labels = np.zeros((args.batch_size, image.shape[0],
image.shape[1]))
mold_image = (image / 255 - mean) / std
if x_pos_map is not None and y_pos_map is not None:
mold_image = np.concatenate((mold_image, x_pos_map, y_pos_map),
-1)
elif gaussian_map is not None:
mold_image = np.concatenate((mold_image, gaussian_map), -1)
batch[batch_index] = mold_image
labels[batch_index] = dataset.load_labels(image_id)
image_ids.append(image_id)
batch_index = batch_index + 1
if batch_index < args.batch_size and idx != data_len - 1:
continue
# ------ end iteration
batch_index = 0
input = batch.transpose((0, 3, 1, 2))
input = torch.from_numpy(input).to(torch.float).to(device)
# get and deal with output
output = model(input)
if type(output) == list:
output = output[0]
if output.size()[-1] < labels.shape[-1]:
output = F.upsample(
output, size=labels.shape[-2:], mode='bilinear')
output = torch.argmax(output, dim=1).cpu().detach().numpy()
# ------ start iteration
input_images = unmold_input(input, True)
for b in range(input_images.shape[0]):
# get data
image_name = os.path.basename(
dataset[image_ids[b]]['image_path'])[:-4]
if args.tform_back:
ori_image = dataset.load_original_image(image_ids[b])
target = dataset.load_original_labels(image_ids[b])
tform_params = dataset.load_align_transform(image_ids[b])
if args.data_settings.find('multi_person') != -1:
ori_shape = (target.shape[0] * 2, target.shape[1] * 2)
ori_image = transform.rescale(ori_image, 2, preserve_range=True).astype(np.uint8)
target = transform.rescale(target, 2, order=0, preserve_range=True).astype(np.uint8)
else:
ori_shape = target.shape[:2]
pred = transform.warp(
output[b],
tform_params,
output_shape=ori_shape,
preserve_range=True)
pred = pred.astype(np.uint8)
else:
ori_image = input_images[b]
target = labels[b].astype(np.uint8)
pred = output[b].astype(np.uint8)
# calculate result
acc_hist_all.collect(target, pred)
acc_hist_single.collect(target, pred)
f1_result = acc_hist_single.get_f1_results()['hair']
print(f'dealing with: input.shape{ori_image.shape} output.shape{pred.shape}')
# visualize result
gt_blended = blend_labels(ori_image, target)
predict_blended = blend_labels(ori_image, pred)
fig, axes = plt.subplots(ncols=2)
axes[0].imshow(predict_blended)
axes[0].set(title=f'predict:%04f' % (f1_result))
axes[1].imshow(gt_blended)
axes[1].set(title='ground-truth')
if args.save:
save_path = os.path.join(save_dir, f'%04f_%s.png' %
(f1_result, image_name))
plt.savefig(save_path)
else:
plt.show()
plt.close(fig)
acc_hist_single.reset()
# ------ end iteration
batch_time.update(time.time() - end)
end = time.time()
image_ids = []
f1_result = acc_hist_all.get_f1_results()['hair']
print('Valiation: [{0}]\t'
'Time {batch_time.val:.3f} ({batch_time.avg:.3f})\t'
'Acc of f-score [{1}]'.format(
len(dataset), f1_result, batch_time=batch_time))
def evaluate_general_dataset(model, dataset):
batch_time = AverageMeter()
acc_hist_all = Acc_score(['hair'])
acc_hist_single = Acc_score(['hair'])
# switch to evaluate mode
model.eval()
with torch.no_grad():
end = time.time()
batch_index = 0
batch = None
labels = None
image_names = []
data_len = len(dataset.image_ids)
for idx, image_id in enumerate(dataset.image_ids):
torch_tensor = dataset[idx]
input, target = torch_tensor['image'].numpy(), torch_tensor[
'label'].numpy()
if batch_index == 0:
batch = np.zeros((args.batch_size, input.shape[0],
input.shape[1], input.shape[2]))
labels = np.zeros((args.batch_size, target.shape[0],
target.shape[1]))
batch[batch_index] = input
labels[batch_index] = target
image_names.append(
os.path.basename(dataset.get_info(idx)['image_path'])[:-4])
batch_index = batch_index + 1
if batch_index < args.batch_size and idx != data_len - 1:
continue
batch_index = 0
input, target = torch.from_numpy(batch).to(
torch.float).to(device), torch.from_numpy(labels).to(
torch.long).to(device)
# get and deal with output
output = model(input)
if type(output) == list: # multi scale output
output = output[0]
print(
f'dealing with: input.shape{input.size()} output.shape{output.size()}'
)
if output.size()[-1] < target.size()[-1]:
output = F.upsample(
output, size=target.size()[-2:], mode='bilinear')
target = target.cpu().detach().numpy()
pred = torch.argmax(output, dim=1).cpu().detach().numpy()
acc_hist_all.collect(target, pred)
acc_hist_single.collect(target, pred)
f1_result = acc_hist_single.get_f1_results()['hair']
input_images = unmold_input(input, keep_dims=True)
for b in range(input_images.shape[0]):
print('deal with', input_images[b].shape, target[b].shape)
gt_blended = blend_labels(input_images[b], target[b])
predict_blended = blend_labels(input_images[b], pred[b])
fig, axes = plt.subplots(ncols=2)
axes[0].imshow(predict_blended)
axes[0].set(title=f'predict:%04f' % (f1_result))
axes[1].imshow(gt_blended)
axes[1].set(title='ground-truth')
if args.save:
save_path = os.path.join(save_dir, f'%04f_%s.png' %
(f1_result, image_names[b]))
plt.savefig(save_path)
else:
plt.show()
plt.close(fig)
acc_hist_single.reset()
batch_time.update(time.time() - end)
end = time.time()
image_names = []
f1_result = acc_hist_all.get_f1_results()['hair']
print('Valiation: [{0}]\t'
'Time {batch_time.val:.3f} ({batch_time.avg:.3f})\t'
'Acc of f-score [{1}]'.format(
len(dataset), f1_result, batch_time=batch_time))
if __name__ == '__main__':
main()