-
Notifications
You must be signed in to change notification settings - Fork 2
/
eval.py
213 lines (171 loc) · 7.6 KB
/
eval.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
import argparse
import numpy as np
import os
os.environ['CUDA_VISIBLE_DEVICES'] = '0'
import torch
import torchvision
import time
from torch.utils.data import DataLoader
import datetime
import cv2
from collections import OrderedDict
import torch.optim
import NYUv2_dataloader as Data
from src.AsymFormer import B0_T
from utils import utils
from utils.utils import load_ckpt, intersectionAndUnion, AverageMeter, accuracy, macc
pth_dir = './model_M1/ckpt_epoch_500.00.pth'
model = B0_T(num_classes=40)
parser = argparse.ArgumentParser(description='RGBD Sementic Segmentation')
parser.add_argument('--data-dir', default='./data', metavar='DIR',
help='path to dataset')
parser.add_argument('-o', '--output', default='./result/', metavar='DIR',
help='path to output')
parser.add_argument('--cuda', action='store_true', default=True,
help='enables CUDA training')
parser.add_argument('--last-ckpt', default='./model/non_local_5173.pth', type=str, metavar='PATH',
help='path to latest checkpoint (default: none)')
parser.add_argument('--num-class', default=40, type=int,
help='number of classes')
parser.add_argument('--visualize', default=False, action='store_true',
help='if output image')
args = parser.parse_args()
image_w = 640
image_h = 480
img_mean = [0.485, 0.456, 0.406]
img_std = [0.229, 0.224, 0.225]
def _load_block_pretrain_weight(model, pretrain_path):
model_dict = model.state_dict()
pretrain_dict = torch.load(pretrain_path)['state_dict']
new_state_dict = OrderedDict()
new_state_dict = {k: v for k, v in pretrain_dict.items() if k in model_dict}
model.load_state_dict(new_state_dict)
# transform
class scaleNorm(object):
def __call__(self, sample):
image, depth, label = sample['image'], sample['depth'], sample['label']
label = label.astype(np.int16)
# Bi-linear
image = cv2.resize(image, (image_w, image_h), cv2.INTER_LINEAR)
# Nearest-neighbor
depth = cv2.resize(depth, (image_w, image_h), cv2.INTER_NEAREST)
label = cv2.resize(label, (image_w, image_h), cv2.INTER_NEAREST)
return {'image': image, 'depth': depth, 'label': label}
class ToTensor(object):
"""Convert ndarrays in sample to Tensors."""
def __call__(self, sample):
image, depth, label = sample['image'], sample['depth'], sample['label']
image = image.transpose((2, 0, 1))
depth = np.expand_dims(depth, 0).astype(np.float64)
return {'image': torch.from_numpy(image).float(),
'depth': torch.from_numpy(depth).float(),
'label': torch.from_numpy(label).float()}
class Normalize(object):
def __call__(self, sample):
image, depth = sample['image'], sample['depth']
origin_image = image.clone()
origin_depth = depth.clone()
image = image / 255
depth = depth / 1000
# image = torchvision.transforms.Normalize(mean=[0.485, 0.456, 0.406],std=[0.229, 0.224, 0.225])(image)
image = torchvision.transforms.Normalize(mean=[0.4850042694973687, 0.41627756261047333, 0.3981809741523051],
std=[0.26415541082494515, 0.2728415392982039, 0.2831175140191598])(
image)
depth = torchvision.transforms.Normalize(mean=[2.8424503515351494], std=[0.9932836506164299])(depth)
sample['origin_image'] = origin_image
sample['origin_depth'] = origin_depth
sample['image'] = image
sample['depth'] = depth
return sample
def visualize_result(img, depth, label, preds, info, args):
# segmentation
img = img.squeeze(0).transpose(0, 2, 1)
dep = depth.squeeze(0).squeeze(0)
dep = (dep * 255 / dep.max()).astype(np.uint8)
dep = cv2.applyColorMap(dep, cv2.COLORMAP_JET)
dep = dep.transpose(2, 1, 0)
seg_color = utils.color_label_eval(label)
# prediction
pred_color = utils.color_label_eval(preds)
# aggregate images and save
im_vis = np.concatenate((img, dep, seg_color, pred_color),
axis=1).astype(np.uint8)
im_vis = im_vis.transpose(2, 1, 0)
img_name = str(info)
# print('write check: ', im_vis.dtype)
cv2.imwrite(os.path.join(args.output,
img_name + '.png'), im_vis)
def time_synchronized():
torch.cuda.synchronize() if torch.cuda.is_available() else None
return time.time()
def inference():
device = torch.device("cuda:0")
_load_block_pretrain_weight(model, pth_dir)
model.eval()
#model._model_deploy()
model.to(device)
val_data = Data.RGBD_Dataset(transform=torchvision.transforms.Compose([scaleNorm(),
ToTensor(),
Normalize()]),
phase_train=False,
data_dir=args.data_dir,
txt_name='test.txt'
)
val_loader = DataLoader(val_data, batch_size=1, shuffle=False, num_workers=0, pin_memory=True)
acc_meter = AverageMeter()
intersection_meter = AverageMeter()
union_meter = AverageMeter()
a_meter = AverageMeter()
b_meter = AverageMeter()
t = 0
acc_collect = []
torch.cuda.synchronize()
starter, ender = torch.cuda.Event(enable_timing=True), torch.cuda.Event(enable_timing=True)
timings = np.zeros((len(val_loader), 1))
dummy_rgb = torch.rand([1, 3, 480, 640], device=device)
dummy_depth = torch.rand([1, 1, 480, 640], device=device)
with torch.no_grad():
for _ in range(10):
_ = model(dummy_rgb, dummy_depth)
for batch_idx, sample in enumerate(val_loader):
origin_image = sample['origin_image'].numpy()
origin_depth = sample['origin_depth'].numpy()
image = sample['image'].to(device)
depth = sample['depth'].to(device)
label = sample['label'].numpy()
starter.record()
pred = model(image, depth)
ender.record()
torch.cuda.synchronize()
curr_time = starter.elapsed_time(ender)
timings[batch_idx] = curr_time
output = torch.max(pred, 1)[1] + 1
output = output.squeeze(0).cpu().numpy()
acc, pix = accuracy(output, label)
acc_collect.append(acc)
intersection, union = intersectionAndUnion(output, label, args.num_class)
acc_meter.update(acc, pix)
a_m, b_m = macc(output, label, args.num_class)
intersection_meter.update(intersection)
union_meter.update(union)
a_meter.update(a_m)
b_meter.update(b_m)
print('[{}] iter {}, accuracy: {}'
.format(datetime.datetime.now().strftime("%Y-%m-%d %H:%M:%S"),
batch_idx, acc))
if args.visualize:
visualize_result(origin_image, origin_depth, label - 1, output - 1, batch_idx, args)
iou = intersection_meter.sum / (union_meter.sum + 1e-10)
for i, _iou in enumerate(iou):
print('class [{}], IoU: {}'.format(i, _iou))
mAcc = (a_meter.average() / (b_meter.average() + 1e-10))
print(mAcc.mean())
print('[Eval Summary]:')
print('Mean IoU: {:.4}, Accuracy: {:.2f}%'
.format(iou.mean(), acc_meter.average() * 100))
print('平均推理时间:', timings.sum() / 654)
np.save('SCC_SRM5', np.array(acc_collect))
if __name__ == '__main__':
if not os.path.exists(args.output):
os.mkdir(args.output)
inference()