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_lossprint('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 = 128def 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_sizedef 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)   -----------------------------