endernewton版本的faster tensorflow實現代碼：
https://github.com/endernewton/tf-faster-rcnn
修改的roi-pooling代碼
# --------------------------------------------------------
# Tensorflow Faster R-CNN
# Licensed under The MIT License [see LICENSE for details]
# Written by Zheqi He and Xinlei Chen
# @ nansbas
# @ 2018.12.30
# --------------------------------------------------------
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
from tensorflow.contrib.slim.python.slim.nets.resnet_v1 import resnet_v1_block
import numpy as np
from nets.network import Network
from model.config import cfg
def resnet_arg_scope(is_training=True,
batch_norm_decay=0.997,
batch_norm_epsilon=1e-5,
batch_norm_scale=True):
batch_norm_params = {
'is_training': False,
'decay': batch_norm_decay,
'epsilon': batch_norm_epsilon,
'scale': batch_norm_scale,
'trainable': False,
'updates_collections': tf.GraphKeys.UPDATE_OPS
}
with arg_scope(
[slim.conv2d],
weights_regularizer=slim.l2_regularizer(cfg.TRAIN.WEIGHT_DECAY),
weights_initializer=slim.variance_scaling_initializer(),
trainable=is_training,
activation_fn=tf.nn.relu,
normalizer_fn=slim.batch_norm,
normalizer_params=batch_norm_params):
with arg_scope([slim.batch_norm], **batch_norm_params) as arg_sc:
return arg_sc
class resnetv1(Network):
def __init__(self, num_layers=50):
Network.__init__(self)
self._feat_stride = [16, ]
self._feat_compress = [1. / float(self._feat_stride[0]), ]
self._num_layers = num_layers
self._scope = 'resnet_v1_%d' % num_layers
self._decide_blocks()
#########Start#################
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") / np.float32(self._feat_stride[0]
y1 = tf.slice(rois, [0, 2], [-1, 1], name="y1") / np.float32(self._feat_stride[0]
x2 = tf.slice(rois, [0, 3], [-1, 1], name="x2") / np.float32(self._feat_stride[0]
y2 = tf.slice(rois, [0, 4], [-1, 1], name="y2") / np.float32(self._feat_stride[0]
# Won't be back-propagated to rois anyway, but to save time
if cfg.RESNET.MAX_POOL:
pre_pool_size = cfg.POOLING_SIZE * 2
spacing_w = (x2 - x1) / pre_pool_size
spacing_h = (y2 - y1) / pre_pool_size
x1 = (x1 + spacing_w / 2 ) / (tf.to_float(bottom_shape[2]) - 1.)
y1 = (y1 + spacing_h / 2 ) / (tf.to_float(bottom_shape[1]) - 1.)
nw = spacing_w * tf.to_float(pre_pool_size - 1 ) / (tf.to_float(bottom_shape[2]) - 1.)
nh = spacing_h * tf.to_float(pre_pool_size - 1 ) / (tf.to_float(bottom_shape[1]) - 1.)
x2 = x1 + nw
y2 = y1 + nh
bboxes=tf.concat([y1, x1, y2, x2], axis=1)
crops = tf.image.crop_and_resize(bottom, bboxes, tf.to_int32(batch_ids), [pre_pool_size, pre_pool_size],
name="crops")
crops = slim.avg_pool2d(crops, [2, 2], [2, 2], padding='SAME')
# crops = slim.max_pool2d(crops, [2, 2], padding='SAME')
else:
pre_pool_size = cfg.POOLING_SIZE
spacing_w = (x2 - x1) / pre_pool_size
spacing_h = (y2 - y1) / pre_pool_size
x1 = (x1 + spacing_w / 2 ) / (tf.to_float(bottom_shape[2]) - 1.)
y1 = (y1 + spacing_h / 2 ) / (tf.to_float(bottom_shape[1]) - 1.)
nw = spacing_w * tf.to_float(pre_pool_size - 1 ) / (tf.to_float(bottom_shape[2]) - 1.)
nh = spacing_h * tf.to_float(pre_pool_size - 1 ) / (tf.to_float(bottom_shape[1]) - 1.)
x2 = x1 + nw
y2 = y1 + nh
bboxes=tf.concat([y1, x1, y2, x2], axis=1)
bboxes = tf.stop_gradient(tf.concat([y1, x1, y2, x2], 1))
crops = tf.image.crop_and_resize(bottom, bboxes, tf.to_int32(batch_ids), [pre_pool_size, pre_pool_size],
name="crops") name="crops")
return crops
#########End#################
# 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._scope, self._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 _image_to_head(self, is_training, reuse=None):
assert (0 <= cfg.RESNET.FIXED_BLOCKS <= 3)
# Now the base is always fixed during training
with slim.arg_scope(resnet_arg_scope(is_training=False)):
net_conv = self._build_base()
if cfg.RESNET.FIXED_BLOCKS > 0:
with slim.arg_scope(resnet_arg_scope(is_training=False)):
net_conv, _ = resnet_v1.resnet_v1(net_conv,
self._blocks[0:cfg.RESNET.FIXED_BLOCKS],
global_pool=False,
include_root_block=False,
reuse=reuse,
scope=self._scope)
if cfg.RESNET.FIXED_BLOCKS < 3:
with slim.arg_scope(resnet_arg_scope(is_training=is_training)):
net_conv, _ = resnet_v1.resnet_v1(net_conv,
self._blocks[cfg.RESNET.FIXED_BLOCKS:-1],
global_pool=False,
include_root_block=False,
reuse=reuse,
scope=self._scope)
self._act_summaries.append(net_conv)
self._layers['head'] = net_conv
# net_conv=slim.layers.conv2d_transpose(net_conv,1024,stride=[2,2],padding="SAME",scope ='deconv')
return net_conv
def _head_to_tail(self, pool5, is_training, reuse=None):
with slim.arg_scope(resnet_arg_scope(is_training=is_training)):
fc7, _ = resnet_v1.resnet_v1(pool5,
self._blocks[-1:],
global_pool=False,
include_root_block=False,
reuse=reuse,
scope=self._scope)
# average pooling done by reduce_mean
fc7 = tf.reduce_mean(fc7, axis=[1, 2])
return fc7
def _decide_blocks(self):
# choose different blocks for different number of layers
if self._num_layers == 50:
self._blocks = [resnet_v1_block('block1', base_depth=64, num_units=3, stride=2),
resnet_v1_block('block2', base_depth=128, num_units=4, stride=2),
# use stride 1 for the last conv4 layer
resnet_v1_block('block3', base_depth=256, num_units=6, stride=1),
resnet_v1_block('block4', base_depth=512, num_units=3, stride=1)]
elif self._num_layers == 101:
self._blocks = [resnet_v1_block('block1', base_depth=64, num_units=3, stride=2),
resnet_v1_block('block2', base_depth=128, num_units=4, stride=2),
# use stride 1 for the last conv4 layer
resnet_v1_block('block3', base_depth=256, num_units=23, stride=1),
resnet_v1_block('block4', base_depth=512, num_units=3, stride=1)]
elif self._num_layers == 152:
self._blocks = [resnet_v1_block('block1', base_depth=64, num_units=3, stride=2),
resnet_v1_block('block2', base_depth=128, num_units=8, stride=2),
# use stride 1 for the last conv4 layer
resnet_v1_block('block3', base_depth=256, num_units=36, stride=1),
resnet_v1_block('block4', base_depth=512, num_units=3, stride=1)]
else:
# other numbers are not supported
raise NotImplementedError
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._scope + '/conv1/weights:0'):
self._variables_to_fix[v.name] = v
continue
if v.name.split(':')[0] in var_keep_dic:
print('Variables 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._scope + "/conv1/weights": conv1_rgb})
restorer_fc.restore(sess, pretrained_model)
sess.run(tf.assign(self._variables_to_fix[self._scope + '/conv1/weights:0'],
tf.reverse(conv1_rgb, [2])))
endernewton已經取消了向上取整，其實已經用了tf.image.corp_and_resize()，通過雙線性差值對非整數座標的情況取值。跟Align描述的區域在於，Align在mask中描述是用如下。
其實計算的是這下圖紅圈的點
而endernewton是計算如下的點。
還是有一點差別的。稍微修改一下就好了。
改成如下：
和mask論文裏面一樣。
最後實驗的時候，先試了pooling的，發現還是直接align雙線性差值得到7x7的效果好一點。
還試了一下下圖這樣的：效果不是很好。
Mask論文裏說到
看來還是mask支路上比較有效果....有誰知道原因麼？