-
Notifications
You must be signed in to change notification settings - Fork 0
/
resnet_v1.py
285 lines (262 loc) · 13.6 KB
/
resnet_v1.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
# --------------------------------------------------------
# Tensorflow Faster R-CNN
# Licensed under The MIT License [see LICENSE for details]
# Written by Zheqi He and Xinlei Chen
# --------------------------------------------------------
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import tensorflow as tf
import tensorflow.contrib.slim as slim
from tensorflow.contrib.slim import losses
from tensorflow.contrib.slim import arg_scope
from tensorflow.contrib.slim.python.slim.nets import resnet_utils
from tensorflow.contrib.slim.python.slim.nets import resnet_v1
import numpy as np
from lib.nets.network import Network
from tensorflow.python.framework import ops
from tensorflow.contrib.layers.python.layers import regularizers
from tensorflow.python.ops import nn_ops
from tensorflow.contrib.layers.python.layers import initializers
from tensorflow.contrib.layers.python.layers import layers
from lib.config import config as cfg
def resnet_arg_scope(is_training=True,
weight_decay=cfg.FLAGS.weight_decay,
batch_norm_decay=0.997,
batch_norm_epsilon=1e-5,
batch_norm_scale=True):
batch_norm_params = {
# NOTE 'is_training' here does not work because inside resnet it gets reset:
# https://github.com/tensorflow/models/blob/master/slim/nets/resnet_v1.py#L187
'is_training': False,
'decay': batch_norm_decay,
'epsilon': batch_norm_epsilon,
'scale': batch_norm_scale,
# 'trainable': cfg.RESNET.BN_TRAIN,
'trainable': False,
'updates_collections': ops.GraphKeys.UPDATE_OPS
}
with arg_scope(
[slim.conv2d],
weights_regularizer=regularizers.l2_regularizer(weight_decay),
weights_initializer=initializers.variance_scaling_initializer(),
trainable=is_training,
activation_fn=nn_ops.relu,
normalizer_fn=layers.batch_norm,
normalizer_params=batch_norm_params):
with arg_scope([layers.batch_norm], **batch_norm_params) as arg_sc:
return arg_sc
class resnetv1(Network):
def __init__(self, batch_size=1, num_layers=50):
Network.__init__(self, batch_size=batch_size)
self._num_layers = num_layers
self._resnet_scope = 'resnet_v1_%d' % num_layers
#self._decide_blocks()
def _crop_pool_layer(self, bottom, rois, name):
with tf.variable_scope(name) as scope:
batch_ids = tf.squeeze(tf.slice(rois, [0, 0], [-1, 1], name="batch_id"), [1])
# Get the normalized coordinates of bboxes
bottom_shape = tf.shape(bottom)
height = (tf.to_float(bottom_shape[1]) - 1.) * np.float32(self._feat_stride[0])
width = (tf.to_float(bottom_shape[2]) - 1.) * np.float32(self._feat_stride[0])
x1 = tf.slice(rois, [0, 1], [-1, 1], name="x1") / width
y1 = tf.slice(rois, [0, 2], [-1, 1], name="y1") / height
x2 = tf.slice(rois, [0, 3], [-1, 1], name="x2") / width
y2 = tf.slice(rois, [0, 4], [-1, 1], name="y2") / height
# Won't be backpropagated to rois anyway, but to save time
bboxes = tf.stop_gradient(tf.concat([y1, x1, y2, x2], 1))
if cfg.FLAGS.max_pool:
pre_pool_size = cfg.FLAGS.roi_pooling_size * 2
crops = tf.image.crop_and_resize(bottom, bboxes, tf.to_int32(batch_ids), [pre_pool_size, pre_pool_size],
name="crops")
crops = slim.max_pool2d(crops, [2, 2], padding='SAME')
else:
crops = tf.image.crop_and_resize(bottom, bboxes, tf.to_int32(batch_ids), [cfg.FLAGS.roi_pooling_size, cfg.FLAGS.roi_pooling_size],
name="crops")
return crops
# Do the first few layers manually, because 'SAME' padding can behave inconsistently
# for images of different sizes: sometimes 0, sometimes 1
def build_base(self):
with tf.variable_scope(self._resnet_scope, self._resnet_scope):
net = resnet_utils.conv2d_same(self._image, 64, 7, stride=2, scope='conv1')
net = tf.pad(net, [[0, 0], [1, 1], [1, 1], [0, 0]])
net = slim.max_pool2d(net, [3, 3], stride=2, padding='VALID', scope='pool1')
return net
def build_network(self, sess, is_training=True):
# select initializers
if cfg.FLAGS.initializer == "truncated":
initializer = tf.truncated_normal_initializer(mean=0.0, stddev=0.01)
initializer_bbox = tf.truncated_normal_initializer(mean=0.0, stddev=0.001)
else:
initializer = tf.random_normal_initializer(mean=0.0, stddev=0.01)
initializer_bbox = tf.random_normal_initializer(mean=0.0, stddev=0.001)
bottleneck = resnet_v1.bottleneck
# choose different blocks for different number of layers
if self._num_layers == 50:
#blocks = [
#resnet_utils.Block('block1', bottleneck,
# [(256, 64, 1)] * 2 + [(256, 64, 2)]),
#resnet_utils.Block('block2', bottleneck,
# [(512, 128, 1)] * 3 + [(512, 128, 2)]),
# Use stride-1 for the last conv4 layer
#resnet_utils.Block('block3', bottleneck,
# [(1024, 256, 1)] * 5 + [(1024, 256, 1)]),
#resnet_utils.Block('block4', bottleneck, [(2048, 512, 1)] * 3)
#]
blocks = [
resnet_v1.resnet_v1_block('block1', base_depth=64, num_units=3, stride=2),
resnet_v1.resnet_v1_block('block2', base_depth=128, num_units=4, stride=2),
resnet_v1.resnet_v1_block('block3', base_depth=256, num_units=6, stride=1),
resnet_v1.resnet_v1_block('block4', base_depth=512, num_units=3, stride=1),
]
elif self._num_layers == 101:
#blocks = [
#resnet_utils.Block('block1', bottleneck,
# [(256, 64, 1)] * 2 + [(256, 64, 2)]),
#resnet_utils.Block('block2', bottleneck,
# [(512, 128, 1)] * 3 + [(512, 128, 2)]),
# Use stride-1 for the last conv4 layer
#resnet_utils.Block('block3', bottleneck,
# [(1024, 256, 1)] * 22 + [(1024, 256, 1)]),
#resnet_utils.Block('block4', bottleneck, [(2048, 512, 1)] * 3)
#]
blocks = [
resnet_v1.resnet_v1_block('block1', base_depth=64, num_units=3, stride=2),
resnet_v1.resnet_v1_block('block2', base_depth=128, num_units=4, stride=2),
resnet_v1.resnet_v1_block('block3', base_depth=256, num_units=23, stride=1),
resnet_v1.resnet_v1_block('block4', base_depth=512, num_units=3, stride=1),
]
elif self._num_layers == 152:
#blocks = [
#resnet_utils.Block('block1', bottleneck,
# [(256, 64, 1)] * 2 + [(256, 64, 2)]),
#resnet_utils.Block('block2', bottleneck,
# [(512, 128, 1)] * 7 + [(512, 128, 2)]),
# Use stride-1 for the last conv4 layer
#resnet_utils.Block('block3', bottleneck,
# [(1024, 256, 1)] * 35 + [(1024, 256, 1)]),
#resnet_utils.Block('block4', bottleneck, [(2048, 512, 1)] * 3)
#]
blocks = [
resnet_v1.resnet_v1_block('block1', base_depth=64, num_units=3, stride=2),
resnet_v1.resnet_v1_block('block2', base_depth=128, num_units=8, stride=2),
resnet_v1.resnet_v1_block('block3', base_depth=256, num_units=36, stride=1),
resnet_v1.resnet_v1_block('block4', base_depth=512, num_units=3, stride=1),
]
else:
# other numbers are not supported
raise NotImplementedError
assert (0 <= cfg.FLAGS.fixed_blocks < 4)
if cfg.FLAGS.fixed_blocks == 3:
with slim.arg_scope(resnet_arg_scope(is_training=False)):
net = self.build_base()
net_conv4, _ = resnet_v1.resnet_v1(net,
blocks[0:cfg.FLAGS.fixed_blocks],
global_pool=False,
include_root_block=False,
scope=self._resnet_scope)
elif cfg.FLAGS.fixed_blocks > 0:
with slim.arg_scope(resnet_arg_scope(is_training=False)):
net = self.build_base()
net, _ = resnet_v1.resnet_v1(net,
blocks[0:cfg.FLAGS.fixed_blocks],
global_pool=False,
include_root_block=False,
scope=self._resnet_scope)
with slim.arg_scope(resnet_arg_scope(is_training=is_training)):
net_conv4, _ = resnet_v1.resnet_v1(net,
blocks[cfg.FLAGS.fixed_blocks:-1],
global_pool=False,
include_root_block=False,
scope=self._resnet_scope)
else: # cfg.FLAGS.fixed_blocks == 0
with slim.arg_scope(resnet_arg_scope(is_training=is_training)):
net = self.build_base()
net_conv4, _ = resnet_v1.resnet_v1(net,
blocks[0:-1],
global_pool=False,
include_root_block=False,
scope=self._resnet_scope)
self._act_summaries.append(net_conv4)
self._layers['head'] = net_conv4
with tf.variable_scope(self._resnet_scope, self._resnet_scope):
# build the anchors for the image
self._anchor_component()
# rpn
rpn = slim.conv2d(net_conv4, 512, [3, 3], trainable=is_training, weights_initializer=initializer,
scope="rpn_conv/3x3")
self._act_summaries.append(rpn)
rpn_cls_score = slim.conv2d(rpn, self._num_anchors * 2, [1, 1], trainable=is_training,
weights_initializer=initializer,
padding='VALID', activation_fn=None, scope='rpn_cls_score')
# change it so that the score has 2 as its channel size
rpn_cls_score_reshape = self._reshape_layer(rpn_cls_score, 2, 'rpn_cls_score_reshape')
rpn_cls_prob_reshape = self._softmax_layer(rpn_cls_score_reshape, "rpn_cls_prob_reshape")
rpn_cls_prob = self._reshape_layer(rpn_cls_prob_reshape, self._num_anchors * 2, "rpn_cls_prob")
rpn_bbox_pred = slim.conv2d(rpn, self._num_anchors * 4, [1, 1], trainable=is_training,
weights_initializer=initializer,
padding='VALID', activation_fn=None, scope='rpn_bbox_pred')
if is_training:
rois, roi_scores = self._proposal_layer(rpn_cls_prob, rpn_bbox_pred, "rois")
rpn_labels = self._anchor_target_layer(rpn_cls_score, "anchor")
# Try to have a determinestic order for the computing graph, for reproducibility
with tf.control_dependencies([rpn_labels]):
rois, _ = self._proposal_target_layer(rois, roi_scores, "rpn_rois")
else:
if cfg.FLAGS.test_mode == 'nms':
rois, _ = self._proposal_layer(rpn_cls_prob, rpn_bbox_pred, "rois")
elif cfg.FLAGS.test_mode == 'top':
rois, _ = self._proposal_top_layer(rpn_cls_prob, rpn_bbox_pred, "rois")
else:
raise NotImplementedError
# rcnn
if cfg.FLAGS.pooling_mode == 'crop':
pool5 = self._crop_pool_layer(net_conv4, rois, "pool5")
else:
raise NotImplementedError
with slim.arg_scope(resnet_arg_scope(is_training=is_training)):
fc7, _ = resnet_v1.resnet_v1(pool5,
blocks[-1:],
global_pool=False,
include_root_block=False,
scope=self._resnet_scope)
with tf.variable_scope(self._resnet_scope, self._resnet_scope):
# Average pooling done by reduce_mean
fc7 = tf.reduce_mean(fc7, axis=[1, 2])
cls_score = slim.fully_connected(fc7, self._num_classes, weights_initializer=initializer,
trainable=is_training, activation_fn=None, scope='cls_score')
cls_prob = self._softmax_layer(cls_score, "cls_prob")
bbox_pred = slim.fully_connected(fc7, self._num_classes * 4, weights_initializer=initializer_bbox,
trainable=is_training,
activation_fn=None, scope='bbox_pred')
self._predictions["rpn_cls_score"] = rpn_cls_score
self._predictions["rpn_cls_score_reshape"] = rpn_cls_score_reshape
self._predictions["rpn_cls_prob"] = rpn_cls_prob
self._predictions["rpn_bbox_pred"] = rpn_bbox_pred
self._predictions["cls_score"] = cls_score
self._predictions["cls_prob"] = cls_prob
self._predictions["bbox_pred"] = bbox_pred
self._predictions["rois"] = rois
self._score_summaries.update(self._predictions)
return rois, cls_prob, bbox_pred
def get_variables_to_restore(self, variables, var_keep_dic):
variables_to_restore = []
for v in variables:
# exclude the first conv layer to swap RGB to BGR
if v.name == (self._resnet_scope + '/conv1/weights:0'):
self._variables_to_fix[v.name] = v
continue
if v.name.split(':')[0] in var_keep_dic:
print('Varibles restored: %s' % v.name)
variables_to_restore.append(v)
return variables_to_restore
def fix_variables(self, sess, pretrained_model):
print('Fix Resnet V1 layers..')
with tf.variable_scope('Fix_Resnet_V1') as scope:
with tf.device("/cpu:0"):
# fix RGB to BGR
conv1_rgb = tf.get_variable("conv1_rgb", [7, 7, 3, 64], trainable=False)
restorer_fc = tf.train.Saver({self._resnet_scope + "/conv1/weights": conv1_rgb})
restorer_fc.restore(sess, pretrained_model)
sess.run(tf.assign(self._variables_to_fix[self._resnet_scope + '/conv1/weights:0'],
tf.reverse(conv1_rgb, [2])))