手写汉字笔迹识别模型：
第一名用的是googleNet，准确率97.3%
def GoogleLeNetSlim(x, num_classes, keep_prob=0.5):
with tf.variable_scope('main'):
t = slim.conv2d(x, 64, [3, 3], [1, 1], padding='SAME', activation_fn=relu, normalizer_fn=slim.batch_norm, scope='conv1')
t = slim.max_pool2d(t, [2, 2], [2, 2], padding='SAME')
t = slim.conv2d(t, 96, [3, 3], [1, 1], padding='SAME', activation_fn=relu, normalizer_fn=slim.batch_norm, scope='conv2')
t = slim.conv2d(t, 192, [3, 3], [1, 1], padding='SAME', activation_fn=relu, normalizer_fn=slim.batch_norm, scope='conv3')
t = slim.max_pool2d(t, [2, 2], [2, 2], padding='SAME')
with tf.variable_scope('block1'):
t = block_slim(t, [64, 96, 128, 16, 32, 32], name='block1') # [?, 16, 16, 256]
with tf.variable_scope('block2'):
t = block_slim(t, [128, 128, 192, 32, 96, 64], name='block1') # [?, 16, 16, 480]
t = tf.nn.max_pool(t, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
with tf.variable_scope('block3'):
t = block_slim(t, [192, 96, 208, 16, 48, 64], name='block1')
t = block_slim(t, [160, 112, 224, 24, 64, 64], name='block2')
t = block_slim(t, [128, 128, 256, 24, 64, 64], name='block3')
t = block_slim(t, [112, 144, 288, 32, 64, 64], name='block4')
t = block_slim(t, [256, 160, 320, 32, 128, 128], name='block5') # [?, 8, 8, 832]
t = tf.nn.max_pool(t, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
with tf.variable_scope('block4'):
t = block_slim(t, [256, 160, 320, 32, 128, 128], name='block1')
t = block_slim(t, [384, 192, 384, 48, 128, 128], name='block2') # [?, 8, 8, 1024]
t = tf.nn.max_pool(t, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
with tf.variable_scope('fc'):
t = slim.flatten(t)
t = slim.fully_connected(slim.dropout(t, keep_prob), 1024, activation_fn=relu, normalizer_fn=slim.batch_norm, scope='fc1')
t = slim.fully_connected(slim.dropout(t, keep_prob), num_classes, activation_fn=None, scope='logits')
return t
TODO：实验下，https://github.com/tflearn/tflearn/blob/master/examples/images/googlenet.py
还有使用inception v3的！！！
def build_graph_all(top_k,scope=None):
keep_prob = tf.placeholder(dtype=tf.float32, shape=[], name='keep_prob')
images = tf.placeholder(dtype=tf.float32, shape=[None, image_size, image_size, 1], name='image_batch')
labels = tf.placeholder(dtype=tf.int64, shape=[None], name='label_batch')
with tf.variable_scope(scope,'Incept_Net',[images]):
with slim.arg_scope([slim.conv2d,slim.max_pool2d,slim.avg_pool2d],stride=1,padding='VALID'):
net = slim.conv2d(images,32,[3,3],scope='conv2d_1a_3x3')
print('tensor 1:' + str(net.get_shape().as_list()))
net = slim.conv2d(net,32,[3,3],scope='conv2d_2a_3x3')
print('tensor 2:' + str(net.get_shape().as_list()))
net = slim.conv2d(net,64,[3,3],padding='SAME',scope='conv2d_2b_3x3')
print('tensor 3:' + str(net.get_shape().as_list()))
net = slim.max_pool2d(net,[3,3],stride=2,scope='maxpool_3a_3x3')
print('tensor 4:' + str(net.get_shape().as_list()))
net = slim.conv2d(net,80,[1,1],scope='conv2d_3b_1x1')
print('tensor 5:' + str(net.get_shape().as_list()))
net = slim.conv2d(net,192,[3,3],scope='conv2d_4a_3x3')
print('tensor 6:' + str(net.get_shape().as_list()))
net = slim.max_pool2d(net,[3,3],stride=2,scope='maxpool_5a_3x3')
print('tensor 7:' + str(net.get_shape().as_list()))
with slim.arg_scope([slim.conv2d,slim.max_pool2d,slim.avg_pool2d],stride=1,padding='SAME'):
with tf.variable_scope('mixed_5b'):
with tf.variable_scope('branch_0'):
branch_0 = slim.conv2d(net,64,[1,1],scope='conv2d_0a_1x1')
with tf.variable_scope('branch_1'):
branch_1 = slim.conv2d(net,48,[1,1],scope='conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1,64,[5,5],scope='conv2d_0b_5x5')
with tf.variable_scope('branch_2'):
branch_2 = slim.conv2d(net,64,[1,1],scope='conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2,96,[3,3],scope='conv2d_0b_3x3')
branch_2 = slim.conv2d(branch_2,96,[3,3],scope='conv2d_0c_3x3')
with tf.variable_scope('branch_3'):
branch_3 = slim.avg_pool2d(net,[3,3],scope='avgpool_0a_3x3')
branch_3 = slim.conv2d(branch_3,32,[1,1],scope='conv2d_0b_1x1')
net = tf.concat([branch_0,branch_1,branch_2,branch_3],3)
print('tensor 8:' + str(net.get_shape().as_list()))
with tf.variable_scope('mixed_5c'):
with tf.variable_scope('branch_0'):
branch_0 = slim.conv2d(net,64,[1,1],scope='conv2d_0a_1x1')
with tf.variable_scope('branch_1'):
branch_1 = slim.conv2d(net,48,[1,1],scope='conv2d_0b_1x1')
branch_1 = slim.conv2d(branch_1,64,[5,5],scope='conv2d_0c_5x5')
with tf.variable_scope('branch_2'):
branch_2 = slim.conv2d(net,64,[1,1],scope='conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2,96,[3,3],scope='conv2d_0b_3x3')
branch_2 = slim.conv2d(branch_2,96,[3,3],scope='conv2d_0c_3x3')
with tf.variable_scope('branch_3'):
branch_3 = slim.avg_pool2d(net,[3,3],scope='avgpool_0a_3x3')
branch_3 = slim.conv2d(branch_3,64,[1,1],scope='conv2d_0b_1x1')
net = tf.concat([branch_0,branch_1,branch_2,branch_3],3)
print('tensor 9:' + str(net.get_shape().as_list()))
with tf.variable_scope('mixed_5d'):
with tf.variable_scope('branch_0'):
branch_0 = slim.conv2d(net,64,[1,1],scope='conv2d_0a_1x1')
with tf.variable_scope('branch_1'):
branch_1 = slim.conv2d(net,48,[1,1],scope='conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1,64,[5,5],scope='conv2d_0b_5x5')
with tf.variable_scope('branch_2'):
branch_2 = slim.conv2d(net,64,[1,1],scope='conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2,96,[3,3],scope='conv2d_0b_3x3')
branch_2 = slim.conv2d(branch_2,96,[3,3],scope='conv2d_0c_3x3')
with tf.variable_scope('branch_3'):
branch_3 = slim.avg_pool2d(net,[3,3],scope='avgpool_0a_3x3')
branch_3 = slim.conv2d(branch_3,64,[1,1],scope='conv2d_0b_1x1')
net = tf.concat([branch_0,branch_1,branch_2,branch_3],3)
print('tensor 10:' + str(net.get_shape().as_list()))
with tf.variable_scope('mixed_6a'):
with tf.variable_scope('branch_0'):
branch_0 = slim.conv2d(net,384,[3,3],stride=2,padding='VALID',scope='conv2d_1a_1x1')
with tf.variable_scope('branch_1'):
branch_1 = slim.conv2d(net,64,[1,1],scope='conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1,96,[3,3],scope='conv2d_0b_3x3')
branch_1 = slim.conv2d(branch_1,96,[3,3],stride=2,padding='VALID',scope='conv2d_1a_1x1')
with tf.variable_scope('branch_2'):
branch_2 = slim.max_pool2d(net,[3,3],stride=2,padding='VALID',scope='maxpool_1a_3x3')
net = tf.concat([branch_0,branch_1,branch_2],3)
print('tensor 11:' + str(net.get_shape().as_list()))
with tf.variable_scope('mixed_6b'):
with tf.variable_scope('branch_0'):
branch_0 = slim.conv2d(net,192,[1,1],scope='conv2d_0a_1x1')
with tf.variable_scope('branch_1'):
branch_1 = slim.conv2d(net,128,[1,1],scope='conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1,128,[1,7],scope='conv2d_0b_1x7')
branch_1 = slim.conv2d(branch_1,192,[7,1],scope='conv2d_0c_7x1')
with tf.variable_scope('branch_2'):
branch_2 = slim.conv2d(net,128,[1,1],scope='conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2,128,[7,1],scope='conv2d_0b_7x1')
branch_2 = slim.conv2d(branch_2,128,[1,7],scope='conv2d_0c_1x7')
branch_2 = slim.conv2d(branch_2,128,[7,1],scope='conv2d_0d_7x1')
branch_2 = slim.conv2d(branch_2,192,[1,7],scope='conv2d_0e_1x7')
with tf.variable_scope('branch_3'):
branch_3 = slim.avg_pool2d(net,[3,3],scope='avgpool_0a_3x3')
branch_3 = slim.conv2d(branch_3,192,[1,1],scope='conv2d_0b_1x1')
net = tf.concat([branch_0,branch_1,branch_2,branch_3],3)
print('tensor 12:' + str(net.get_shape().as_list()))
with tf.variable_scope('mixed_6c'):
with tf.variable_scope('branch_0'):
branch_0 = slim.conv2d(net,192,[1,1],scope='conv2d_0a_1x1')
with tf.variable_scope('branch_1'):
branch_1 = slim.conv2d(net,160,[1,1],scope='conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1,160,[1,7],scope='conv2d_0b_1x7')
branch_1 = slim.conv2d(branch_1,192,[7,1],scope='conv2d_0c_7x1')
with tf.variable_scope('branch_2'):
branch_2 = slim.conv2d(net,160,[1,1],scope='conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2,160,[7,1],scope='conv2d_0b_7x1')
branch_2 = slim.conv2d(branch_2,160,[1,7],scope='conv2d_0c_1x7')
branch_2 = slim.conv2d(branch_2,160,[7,1],scope='conv2d_0d_7x1')
branch_2 = slim.conv2d(branch_2,192,[1,7],scope='conv2d_0e_1x7')
with tf.variable_scope('branch_3'):
branch_3 = slim.avg_pool2d(net,[3,3],scope='avgpool_0a_3x3')
branch_3 = slim.conv2d(branch_3,192,[1,1],scope='conv2d_0b_1x1')
net = tf.concat([branch_0,branch_1,branch_2,branch_3],3)
print('tensor 13:' + str(net.get_shape().as_list()))
with tf.variable_scope('mixed_6d'):
with tf.variable_scope('branch_0'):
branch_0 = slim.conv2d(net,192,[1,1],scope='conv2d_0a_1x1')
with tf.variable_scope('branch_1'):
branch_1 = slim.conv2d(net,160,[1,1],scope='conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1,160,[1,7],scope='conv2d_0b_1x7')
branch_1 = slim.conv2d(branch_1,192,[7,1],scope='conv2d_0c_7x1')
with tf.variable_scope('branch_2'):
branch_2 = slim.conv2d(net,160,[1,1],scope='conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2,160,[7,1],scope='conv2d_0b_7x1')
branch_2 = slim.conv2d(branch_2,160,[1,7],scope='conv2d_0c_1x7')
branch_2 = slim.conv2d(branch_2,160,[7,1],scope='conv2d_0d_7x1')
branch_2 = slim.conv2d(branch_2,192,[1,7],scope='conv2d_0e_1x7')
with tf.variable_scope('branch_3'):
branch_3 = slim.avg_pool2d(net,[3,3],scope='avgpool_0a_3x3')
branch_3 = slim.conv2d(branch_3,192,[1,1],scope='conv2d_0b_1x1')
net = tf.concat([branch_0,branch_1,branch_2,branch_3],3)
print('tensor 14:' + str(net.get_shape().as_list()))
with tf.variable_scope('mixed_6e'):
with tf.variable_scope('branch_0'):
branch_0 = slim.conv2d(net,192,[1,1],scope='conv2d_0a_1x1')
with tf.variable_scope('branch_1'):
branch_1 = slim.conv2d(net,192,[1,1],scope='conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1,192,[1,7],scope='conv2d_0b_1x7')
branch_1 = slim.conv2d(branch_1,192,[7,1],scope='conv2d_0c_7x1')
with tf.variable_scope('branch_2'):
branch_2 = slim.conv2d(net,192,[1,1],scope='conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2,192,[7,1],scope='conv2d_0b_7x1')
branch_2 = slim.conv2d(branch_2,192,[1,7],scope='conv2d_0c_1x7')
branch_2 = slim.conv2d(branch_2,192,[7,1],scope='conv2d_0d_7x1')
branch_2 = slim.conv2d(branch_2,192,[1,7],scope='conv2d_0e_1x7')
with tf.variable_scope('branch_3'):
branch_3 = slim.avg_pool2d(net,[3,3],scope='avgpool_0a_3x3')
branch_3 = slim.conv2d(branch_3,192,[1,1],scope='conv2d_0b_1x1')
net = tf.concat([branch_0,branch_1,branch_2,branch_3],3)
print('tensor 15:' + str(net.get_shape().as_list()))
with tf.variable_scope('mixed_7a'):
with tf.variable_scope('branch_0'):
branch_0 = slim.conv2d(net,192,[1,1],scope='conv2d_0a_1x1')
branch_0 = slim.conv2d(branch_0,320,[3,3],stride=2,padding='VALID',scope='conv2d_1a_3x3')
with tf.variable_scope('branch_1'):
branch_1 = slim.conv2d(net,192,[1,1],scope='conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1,192,[1,7],scope='conv2d_0b_1x7')
branch_1 = slim.conv2d(branch_1,192,[7,1],scope='conv2d_0c_7x1')
branch_1 = slim.conv2d(branch_1,192,[3,3],stride=2,padding='VALID',scope='conv2d_1a_3x3')
with tf.variable_scope('branch_2'):
branch_2 = slim.max_pool2d(net,[3,3],stride=2,padding='VALID',scope='maxpool_1a_3x3')
net = tf.concat([branch_0,branch_1,branch_2],3)
print('tensor 16:' + str(net.get_shape().as_list()))
with tf.variable_scope('mixed_7b'):
with tf.variable_scope('branch_0'):
branch_0 = slim.conv2d(net,320,[1,1],scope='conv2d_0a_1x1')
with tf.variable_scope('branch_1'):
branch_1 = slim.conv2d(net,384,[1,1],scope='conv2d_0a_1x1')
branch_1 = tf.concat([
slim.conv2d(branch_1,384,[1,3],scope='conv2d_0b_1x3'),
slim.conv2d(branch_1,384,[3,1],scope='conv2d_0b_3x1')
],3)
with tf.variable_scope('branch_2'):
branch_2 = slim.conv2d(net,448,[1,1],scope='conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2,384,[3,3],scope='conv2d_0b_3x3')
branch_2 = tf.concat([
slim.conv2d(branch_2,384,[1,3],scope='conv2d_0c_1x3'),
slim.conv2d(branch_2,384,[3,1],scope='conv2d_0d_3x1')
],3)
with tf.variable_scope('branch_3'):
branch_3 = slim.avg_pool2d(net,[3,3],scope='avgpool_0a_3x3')
branch_3 = slim.conv2d(branch_3,192,[1,1],scope='conv2d_0b_1x1')
net = tf.concat([branch_0,branch_1,branch_2,branch_3],3)
print('tensor 17:' + str(net.get_shape().as_list()))
with tf.variable_scope('mixed_7c'):
with tf.variable_scope('branch_0'):
branch_0 = slim.conv2d(net,320,[1,1],scope='conv2d_0a_1x1')
with tf.variable_scope('branch_1'):
branch_1 = slim.conv2d(net,384,[1,1],scope='conv2d_0a_1x1')
branch_1 = tf.concat([
slim.conv2d(branch_1,384,[1,3],scope='conv2d_0b_1x3'),
slim.conv2d(branch_1,384,[3,1],scope='conv2d_0c_3x1')],3)
with tf.variable_scope('branch_2'):
branch_2 = slim.conv2d(net,448,[1,1],scope='conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2,384,[3,3],scope='conv2d_0b_3x3')
branch_2 = tf.concat([
slim.conv2d(branch_2,384,[1,3],scope='conv2d_0c_1x3'),
slim.conv2d(branch_2,384,[3,1],scope='conv2d_0d_3x1')],3)
with tf.variable_scope('branch_3'):
branch_3 = slim.avg_pool2d(net,[3,3],scope='avgpool_0a_3x3')
branch_3 = slim.conv2d(branch_3,192,[1,1],scope='conv2d_0b_1x1')
net = tf.concat([branch_0,branch_1,branch_2,branch_3],3)
print('tensor 18:' + str(net.get_shape().as_list()))
with slim.arg_scope([slim.conv2d,slim.max_pool2d,slim.avg_pool2d],stride=1,padding='SAME'):
with tf.variable_scope('logits'):
net = slim.avg_pool2d(net,[3,3],padding='VALID',scope='avgpool_1a_3x3')
print('tensor 19:' + str(net.get_shape().as_list()))
net = slim.dropout(net,keep_prob=keep_prob,scope='dropout_1b')
logits = slim.conv2d(net, char_size,[2,2],padding='VALID',activation_fn=None,normalizer_fn=None,
scope='conv2d_1c_2x2')
print('logits 1:' + str(logits.get_shape().as_list()))
logits = tf.squeeze(logits,[1,2],name='spatialsqueeze')
print('logits 2:' + str(logits.get_shape().as_list()))
regularization_loss = tf.reduce_sum(tf.get_collection(tf.GraphKeys.REGULARIZATION_LOSSES))
loss = tf.reduce_mean(tf.nn.sparse_softmax_cross_entropy_with_logits(logits=logits, labels=labels))
total_loss = loss + regularization_loss
print('get total_loss')
accuracy = tf.reduce_mean(tf.cast(tf.equal(tf.argmax(logits, 1), labels), tf.float32))
global_step = tf.get_variable("step", [], initializer=tf.constant_initializer(0.0), trainable=False)
rate = tf.train.exponential_decay(2e-3, global_step, decay_steps=2000, decay_rate=0.97, staircase=True)
update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS)
with tf.control_dependencies(update_ops):
train_op = tf.train.AdamOptimizer(learning_rate=rate).minimize(total_loss, global_step=global_step)
probabilities = tf.nn.softmax(logits)
tf.summary.scalar('loss', loss)
tf.summary.scalar('accuracy', accuracy)
merged_summary_op = tf.summary.merge_all()
predicted_val_top_k, predicted_index_top_k = tf.nn.top_k(probabilities, k=top_k)
accuracy_in_top_k = tf.reduce_mean(tf.cast(tf.nn.in_top_k(probabilities, labels, top_k), tf.float32))
return {'images': images,
'labels': labels,
'keep_prob': keep_prob,
'top_k': top_k,
'global_step': global_step,
'train_op': train_op,
'loss': total_loss,
'accuracy': accuracy,
'accuracy_top_k': accuracy_in_top_k,
'merged_summary_op': merged_summary_op,
'predicted_distribution': probabilities,
'predicted_index_top_k': predicted_index_top_k,
'predicted_val_top_k': predicted_val_top_k}
用resnet v2的:
resnet_v2.default_image_size = 128
def resnet_v2_50(inputs,
num_classes=None,
is_training=True,
global_pool=True,
output_stride=None,
spatial_squeeze=True,
reuse=None,
scope='resnet_v2_50'):
"""ResNet-50 model of [1]. See resnet_v2() for arg and return description."""
blocks = [
resnet_v2_block('block1', base_depth=64, num_units=3, stride=2),
resnet_v2_block('block2', base_depth=128, num_units=4, stride=2),
resnet_v2_block('block3', base_depth=256, num_units=6, stride=2),
resnet_v2_block('block4', base_depth=512, num_units=3, stride=1),
]
return resnet_v2(inputs, blocks, num_classes, is_training=is_training,
global_pool=global_pool, output_stride=output_stride,
include_root_block=True, spatial_squeeze=spatial_squeeze,
reuse=reuse, scope=scope)
resnet_v2_50.default_image_size = resnet_v2.default_image_size
def resnet_v2_101(inputs,
num_classes=None,
is_training=True,
global_pool=True,
output_stride=None,
spatial_squeeze=True,
reuse=None,
scope='resnet_v2_101'):
"""ResNet-101 model of [1]. See resnet_v2() for arg and return description."""
blocks = [
resnet_v2_block('block1', base_depth=64, num_units=3, stride=2),
resnet_v2_block('block2', base_depth=128, num_units=4, stride=2),
resnet_v2_block('block3', base_depth=256, num_units=23, stride=2),
resnet_v2_block('block4', base_depth=512, num_units=3, stride=1),
]
return resnet_v2(inputs, blocks, num_classes, is_training=is_training,
global_pool=global_pool, output_stride=output_stride,
include_root_block=True, spatial_squeeze=spatial_squeeze,
reuse=reuse, scope=scope)
def build_graph(top_k, is_training):
# with tf.device('/cpu:0'):
keep_prob = tf.placeholder(dtype=tf.float32, shape=[], name='keep_prob')
images = tf.placeholder(dtype=tf.float32, shape=[None, 128, 128, 1], name='image_batch')
labels = tf.placeholder(dtype=tf.int64, shape=[None], name='label_batch')
logits, _ = resnet_v2_50(images, num_classes=3755, is_training=is_training, global_pool=True,
output_stride=None, spatial_squeeze=True, reuse=None)
-----------------------------
