-
Notifications
You must be signed in to change notification settings - Fork 0
/
predict_keras_datasets.py
159 lines (130 loc) · 6.25 KB
/
predict_keras_datasets.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
import os
import sys
import argparse
import configparser
from datetime import datetime
import pytz
from distutils.util import strtobool
import numpy as np
from core.layers import PixelCNN
from core.utils import Utils
from keras.datasets import mnist
from keras.utils import np_utils
from keras.utils.visualize_util import plot
def predict(argv=None):
''' generate image from conditional Gated PixelCNN model
Usage:
python predict_keras_datasets.py -c sample_predict.cfg : prediction example using configfile
python predict_keras_datasets.py --option1 hoge ... : predict with command-line options
python predict_keras_datasets.py -c predict.cfg --opt1 hoge... : overwrite config options with command-line options
'''
### parsing arguments from command-line or config-file ###
if argv is None:
argv = sys.argv
conf_parser = argparse.ArgumentParser(
description=__doc__, # printed with -h/--help
formatter_class=argparse.RawDescriptionHelpFormatter,
add_help=False
)
conf_parser.add_argument("-c", "--conf_file",
help="Specify config file", metavar="FILE_PATH")
args, remaining_argv = conf_parser.parse_known_args()
defaults = {}
if args.conf_file:
config = configparser.SafeConfigParser()
config.read([args.conf_file])
defaults.update(dict(config.items("General")))
parser = argparse.ArgumentParser(
parents=[conf_parser]
)
parser.set_defaults(**defaults)
parser.add_argument("--checkpoint_file", help="Checkpoint file [Required]", type=str, metavar="FILE_PATH")
parser.add_argument("--conditional", help="model the conditional distribution p(x|h) (default:False)", type=str, metavar="BOOL")
parser.add_argument("--dataset_name", help="{'mnist','cifar10','cifar100'}", type=str, metavar="DATASET_NAME")
parser.add_argument("--class_label", help="Class label (default: 0)",type=int, metavar="INT")
parser.add_argument("--nb_images", help="Number of images to generate",type=int, metavar="INT")
parser.add_argument("--batch_size", help="Batch size at prediction",type=int, metavar="INT")
parser.add_argument("--temperature", help="Temparature value for sampling diverse values (default: 1.0)",type=float, metavar="FLOAT")
parser.add_argument("--nb_pixelcnn_layers", help="Number of PixelCNN Layers (except last two ReLu layers)",type=int, metavar="INT")
parser.add_argument("--nb_filters", help="Number of filters for each layer",type=int, metavar="INT")
parser.add_argument("--filter_size_1st", help="Filter size for the first layer. (default: (7,7))", metavar="INT,INT")
parser.add_argument("--filter_size", help="Filter size for the subsequent layers. (default: (3,3))", metavar="INT,INT")
parser.add_argument("--save_path", help="Root directory which generated images are saved (default: /tmp/pixelcnn/results)", type=str, metavar="DIR_PATH")
args = parser.parse_args(remaining_argv)
utils = Utils()
try:
checkpoint_file = args.checkpoint_file
dataset_name = args.dataset_name
except ValueError:
sys.exit("Error: --checkpoint_file must be specified.")
conditional = strtobool(args.conditional) if args.conditional else False
temperature = args.temperature if args.temperature else 1.0
### mnist image size ###
if dataset_name == 'mnist':
input_size = (28, 28)
nb_classes = 10
nb_channels = 1
elif dataset_name == 'cifar10':
input_size = (32, 32)
nb_classes = 10
nb_channels = 3
elif dataset_name == 'cifar100':
input_size = (32, 32)
nb_classes = 100
nb_channels = 3
### build PixelCNN model ###
model_params = {}
model_params['input_size'] = input_size
model_params['nb_channels'] = nb_channels
model_params['conditional'] = conditional
model_params['latent_dim'] = nb_classes
if args.nb_pixelcnn_layers:
model_params['nb_pixelcnn_layers'] = int(args.nb_pixelcnn_layers)
if args.nb_filters:
model_params['nb_filters'] = int(args.nb_filters)
if args.filter_size_1st:
model_params['filter_size_1st'] = tuple(map(int, args.filter_size_1st.split(',')))
if args.filter_size:
model_params['filter_size'] = tuple(map(int, args.filter_size.split(',')))
save_path = args.save_path if args.save_path else '/tmp/pixelcnn/results'
if not os.path.exists(save_path):
os.makedirs(save_path)
pixelcnn = PixelCNN(**model_params)
pixelcnn.build_model()
pixelcnn.model.load_weights(checkpoint_file)
## prepare zeros array
class_label = int(args.class_label) if args.class_label else 0
nb_images = int(args.nb_images) if args.nb_images else 8
batch_size = int(args.batch_size) if args.batch_size else nb_images
if dataset_name == 'mnist':
X_pred = np.zeros((nb_images, input_size[0], input_size[1], 1))
else:
X_pred = np.zeros((nb_images, input_size[0], input_size[1], 3))
if conditional:
h_pred = np_utils.to_categorical(class_label, nb_classes)
h_pred = np.repeat(h_pred, nb_images, axis=0)
### generate images pixel by pixel
for i in range(input_size[0]):
for j in range(input_size[1]):
for k in range(nb_channels):
if conditional:
x = [X_pred, h_pred]
else:
x = X_pred
next_X_pred = pixelcnn.model.predict(x, batch_size)
if dataset_name == 'mnist':
binarizer = lambda x: utils.binarize_val(x)
binarized_pred = np.vectorize(binarizer)(next_X_pred[:,i,j,k])
X_pred[:,i,j,k] = binarized_pred
else:
sampled_pred = next_X_pred[:,i*input_size[1]*nb_channels+j*nb_channels+k,:]
sampled_pred = np.array([utils.sample(sampled_pred[i]) for i in range(len(sampled_pred))])
X_pred[:,i,j,k] = sampled_pred
X_pred = np.squeeze(X_pred)
X_pred = (255*X_pred).astype(np.uint8)
### save images ###
for i in range(nb_images):
utils.save_generated_image(X_pred[i], 'generated_'+str(i)+'.jpg', save_path)
return (0)
if __name__ == '__main__':
sys.exit(predict())