-
Notifications
You must be signed in to change notification settings - Fork 28
/
monodepth_dataloader.py
149 lines (118 loc) · 7.47 KB
/
monodepth_dataloader.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
# Copyright UCL Business plc 2017. Patent Pending. All rights reserved.
#
# The MonoDepth Software is licensed under the terms of the UCLB ACP-A licence
# which allows for non-commercial use only, the full terms of which are made
# available in the LICENSE file.
#
# For any other use of the software not covered by the UCLB ACP-A Licence,
# please contact info@uclb.com
"""Monodepth data loader.
"""
from __future__ import absolute_import, division, print_function
import tensorflow as tf
def string_length_tf(t):
return tf.py_func(len, [t], [tf.int64])
class MonodepthDataloader(object):
"""monodepth dataloader"""
def __init__(self, data_path, filenames_file, params, dataset, mode, noShuffle=False):
self.data_path = data_path
self.params = params
self.dataset = dataset
self.mode = mode
self.left_image_batch = None
self.right_image_batch = None
self.semantic_image_batch = None
self.valid_image_batch = None
input_queue = tf.train.string_input_producer([filenames_file], shuffle=False)
line_reader = tf.TextLineReader()
_, line = line_reader.read(input_queue)
split_line = tf.string_split([line]).values
# we load only one image for test, except if we trained a stereo model
if mode == 'test' and not self.params.do_stereo:
left_image_path = tf.string_join([self.data_path, split_line[0]])
left_image_o = self.read_image(left_image_path)
# semantic_image_path = tf.string_join([self.data_path, split_line[2]])
# semantic_image_o = self.read_semantic_gt(semantic_image_path)
else:
left_image_path = tf.string_join([self.data_path, split_line[0]])
right_image_path = tf.string_join([self.data_path, split_line[1]])
semantic_image_path = tf.string_join([self.data_path, split_line[2]])
left_image_o = self.read_image(left_image_path)
right_image_o = self.read_image(right_image_path)
semantic_image_o, valid_image_o = self.read_semantic_gt(semantic_image_path)
if mode == 'train':
# randomly flip images
do_flip = tf.random_uniform([], 0, 1)
left_image = tf.cond(do_flip > 0.5, lambda: tf.image.flip_left_right(right_image_o), lambda: left_image_o)
right_image = tf.cond(do_flip > 0.5, lambda: tf.image.flip_left_right(left_image_o), lambda: right_image_o)
semantic_image = semantic_image_o #tf.cond(do_flip > 0.5, lambda: tf.image.flip_left_right(semantic_image_o), lambda: semantic_image_o)
valid_image = tf.cond(do_flip > 0.5, lambda: tf.zeros([self.params.height, self.params.width, 1], tf.float32), lambda: valid_image_o)
# randomly augment images
do_augment = tf.random_uniform([], 0, 1)
left_image, right_image = tf.cond(do_augment > 0.5, lambda: self.augment_image_pair(left_image, right_image), lambda: (left_image, right_image))
left_image.set_shape( [None, None, 3])
right_image.set_shape([None, None, 3])
semantic_image.set_shape([None, None, 1])
valid_image.set_shape([None, None, 1])
# capacity = min_after_dequeue + (num_threads + a small safety margin) * batch_size
min_after_dequeue = 2048
capacity = min_after_dequeue + 4 * params.batch_size
if noShuffle:
self.left_image_batch, self.right_image_batch, self.semantic_image_batch, self.valid_image_batch = tf.train.batch([left_image, right_image, semantic_image, valid_image], params.batch_size, capacity)
else:
self.left_image_batch, self.right_image_batch, self.semantic_image_batch, self.valid_image_batch = tf.train.shuffle_batch([left_image, right_image, semantic_image, valid_image], params.batch_size, capacity, min_after_dequeue, params.num_threads)
elif mode == 'test':
self.left_image_batch = tf.stack([left_image_o, tf.image.flip_left_right(left_image_o)], 0)
self.left_image_batch.set_shape([2, None, None, 3])
# self.semantic_image_batch = tf.stack([semantic_image_o, tf.image.flip_left_right(semantic_image_o)], 0)
# self.semantic_image_batch.set_shape( [2, None, None, 1])
if self.params.do_stereo:
self.right_image_batch = tf.stack([right_image_o, tf.image.flip_left_right(right_image_o)], 0)
self.right_image_batch.set_shape( [2, None, None, 3])
def augment_image_pair(self, left_image, right_image):
# randomly shift gamma
random_gamma = tf.random_uniform([], 0.8, 1.2)
left_image_aug = left_image ** random_gamma
right_image_aug = right_image ** random_gamma
# randomly shift brightness
random_brightness = tf.random_uniform([], 0.5, 2.0)
left_image_aug = left_image_aug * random_brightness
right_image_aug = right_image_aug * random_brightness
# randomly shift color
random_colors = tf.random_uniform([3], 0.8, 1.2)
white = tf.ones([tf.shape(left_image)[0], tf.shape(left_image)[1]])
color_image = tf.stack([white * random_colors[i] for i in range(3)], axis=2)
left_image_aug *= color_image
right_image_aug *= color_image
# saturate
left_image_aug = tf.clip_by_value(left_image_aug, 0, 1)
right_image_aug = tf.clip_by_value(right_image_aug, 0, 1)
return left_image_aug, right_image_aug
def read_semantic_gt(self, image_path):
# tf.decode_image does not return the image size, this is an ugly workaround to handle both jpeg and png
path_length = string_length_tf(image_path)[0]
file_extension = tf.substr(image_path, path_length - 3, 3)
file_cond = tf.equal(file_extension, 'png')
image = tf.cond(file_cond, lambda: tf.image.decode_png(tf.read_file(image_path)), lambda: tf.zeros([self.params.height, self.params.width, 1], tf.uint8))
# if the dataset is cityscapes, we crop the last fifth to remove the car hood
if self.dataset == 'cityscapes':
o_height = tf.shape(image)[0]
crop_height = (o_height * 4) // 5
image = image[:crop_height,:,:]
image = tf.to_int32(tf.image.resize_images(image, [self.params.height, self.params.width], tf.image.ResizeMethod.NEAREST_NEIGHBOR))
valid = tf.cond(file_cond, lambda: tf.ones([self.params.height, self.params.width, 1], tf.float32), lambda: tf.zeros([self.params.height, self.params.width, 1], tf.float32))
return image, valid
def read_image(self, image_path):
# tf.decode_image does not return the image size, this is an ugly workaround to handle both jpeg and png
path_length = string_length_tf(image_path)[0]
file_extension = tf.substr(image_path, path_length - 3, 3)
file_cond = tf.equal(file_extension, 'jpg')
image = tf.cond(file_cond, lambda: tf.image.decode_jpeg(tf.read_file(image_path)), lambda: tf.image.decode_png(tf.read_file(image_path)))
# if the dataset is cityscapes, we crop the last fifth to remove the car hood
if self.dataset == 'cityscapes':
o_height = tf.shape(image)[0]
crop_height = (o_height * 4) // 5
image = image[:crop_height,:,:]
image = tf.image.convert_image_dtype(image, tf.float32)
image = tf.image.resize_images(image, [self.params.height, self.params.width], tf.image.ResizeMethod.AREA)
return image