-
Notifications
You must be signed in to change notification settings - Fork 3
/
inference_rw.py
192 lines (149 loc) · 6.71 KB
/
inference_rw.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
# Copyright (C) 2020 * Ltd. All rights reserved.
# author : Sanghyeon Jo <josanghyeokn@gmail.com>
import os
import sys
import copy
import shutil
import random
import argparse
import numpy as np
import math
from tqdm import tqdm
import imageio
import torch
import torch.nn as nn
import torch.nn.functional as F
from torchvision import transforms
from torch.utils.tensorboard import SummaryWriter
from torch.utils.data import DataLoader
from core.puzzle_utils import *
from core.networks import *
from core.datasets import *
from tools.general.io_utils import *
from tools.general.time_utils import *
from tools.general.json_utils import *
from tools.ai.log_utils import *
from tools.ai.demo_utils import *
from tools.ai.optim_utils import *
from tools.ai.torch_utils import *
from tools.ai.evaluate_utils import *
from tools.ai.augment_utils import *
from tools.ai.randaugment import *
parser = argparse.ArgumentParser()
###############################################################################
# Dataset
###############################################################################
parser.add_argument('--seed', default=0, type=int)
parser.add_argument('--num_workers', default=4, type=int)
parser.add_argument('--data_dir', default='../VOC2012/', type=str)
parser.add_argument('--start', default=0.0, type=float)
parser.add_argument('--end', default=1.0, type=float)
###############################################################################
# Network
###############################################################################
parser.add_argument('--architecture', default='resnet50', type=str)
###############################################################################
# Inference parameters
###############################################################################
parser.add_argument('--model_name', default='', type=str)
parser.add_argument('--cam_dir', default='', type=str)
parser.add_argument('--domain', default='train', type=str)
parser.add_argument('--beta', default=10, type=int)
parser.add_argument('--exp_times', default=8, type=int)
parser.add_argument('--image_size', default=512, type=int)
parser.add_argument('--clear_cache', default=False, type=str2bool)
if __name__ == '__main__':
###################################################################################
# Arguments
###################################################################################
args = parser.parse_args()
experiment_name = args.model_name
if 'train' in args.domain:
experiment_name += '@train'
else:
experiment_name += '@val'
experiment_name += '@beta=%d'%args.beta
experiment_name += '@exp_times=%d'%args.exp_times
experiment_name += '@rw'
cam_dir = f'./experiments/predictions/{args.cam_dir}/'
pred_dir = create_directory(f'./experiments/predictions/{experiment_name}/')
model_path = './experiments/models/' + f'{args.model_name}.pth'
set_seed(args.seed)
log_func = lambda string='': print(string)
###################################################################################
# Transform, Dataset, DataLoader
###################################################################################
imagenet_mean = [0.485, 0.456, 0.406]
imagenet_std = [0.229, 0.224, 0.225]
normalize_fn = Normalize(imagenet_mean, imagenet_std)
meta_dic = read_json('./data/VOC_2012.json')
if args.domain == 'test':
dataset = VOC_Dataset_For_Evaluation(args.data_dir, args.domain)
else:
dataset = VOC_Dataset_For_Making_CAM(args.data_dir, args.domain)
###################################################################################
# Network
###################################################################################
stride = 4
path_index = PathIndex(radius=10, default_size=(args.image_size // stride, args.image_size // stride))
model = AffinityNet(args.architecture, path_index)
model = model.cuda()
model.eval()
log_func('[i] Architecture is {}'.format(args.architecture))
log_func('[i] Total Params: %.2fM'%(calculate_parameters(model)))
log_func()
try:
use_gpu = os.environ['CUDA_VISIBLE_DEVICES']
except KeyError:
use_gpu = '0'
the_number_of_gpu = len(use_gpu.split(','))
if the_number_of_gpu > 1:
log_func('[i] the number of gpu : {}'.format(the_number_of_gpu))
model = nn.DataParallel(model)
load_model(model, model_path, parallel=the_number_of_gpu > 1)
#################################################################################################
# Evaluation
#################################################################################################
eval_timer = Timer()
print('rw output dir: {}'.format(pred_dir))
print('total number: {}'.format(len(dataset)))
with torch.no_grad():
dataset_len = len(dataset)
start = int(dataset_len * args.start)
end = int(dataset_len * args.end)
length = end - start
for item_id in tqdm(
range(start, end),
total=length,
dynamic_ncols=True,
):
item = dataset.__getitem__(item_id)
if args.domain == 'test':
ori_image, image_id, gt_mask = item # (gt_mask is None)
else:
ori_image, image_id, label, gt_mask = item
ori_w, ori_h = ori_image.size
npy_path = pred_dir + image_id + '.npy'
if os.path.isfile(npy_path) and (not args.clear_cache):
continue
# preprocessing
image = np.asarray(ori_image)
image = normalize_fn(image)
image = image.transpose((2, 0, 1))
image = torch.from_numpy(image)
flipped_image = image.flip(-1)
images = torch.stack([image, flipped_image])
images = images.cuda()
edge = model.get_edge(images, image_size=(512, 512), stride=stride)
# postprocessing
cam_dict = np.load(cam_dir + image_id + '.npy', allow_pickle=True).item()
cams = cam_dict['cam']
if isinstance(cams, np.ndarray):
cams = torch.from_numpy(cams)
cam_downsized_values = cams.cuda()
rw = propagate_to_edge(cam_downsized_values, edge, beta=args.beta, exp_times=args.exp_times, radius=5)
rw_up = F.interpolate(rw, scale_factor=stride, mode='bilinear', align_corners=False)[..., 0, :ori_h, :ori_w]
rw_up = rw_up / torch.max(rw_up)
np.save(npy_path, {"keys": cam_dict['keys'], "rw": rw_up.cpu().numpy()})
print()
print("python evaluate.py --experiment_name {} --domain {}".format(experiment_name, args.domain))