TFLite:使用1维CNN处理序列数据的过程

2024-06-03 14:58

本文主要是介绍TFLite:使用1维CNN处理序列数据的过程,希望对大家解决编程问题提供一定的参考价值,需要的开发者们随着小编来一起学习吧!

开发环境

tf.__version__
'2.0.0-beta1'

tf.keras.__version__
'2.2.4-tf'

数据来源

http://www.cis.fordham.edu/wisdm/dataset.php

根据sensor数据x, y, z分类出Downstairs, Upstairs, jogging, sitting, standing, walking 6个类别

数据预处理函数

import pandas as pd
import numpy as np

import matplotlib.pyplot as plt
from scipy import stats
from sklearn import metrics
from sklearn.metrics import classification_report
from sklearn import preprocessing

import keras
import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Dropout, Flatten, Reshape, GlobalAveragePooling1D
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Conv1D, MaxPooling1D


columns = ['user','activity','timestamp', 'x-axis', 'y-axis', 'z-axis']
def read_data(file_patch):
    train = pd.read_csv(file_patch, header = None, names = columns)
    train = train.dropna()
    return train

# mean and std
def feature_normalize(feature):
    mu = np.mean(feature, axis=0)
    sigma = np.std(feature, axis=0)
    return (feature - mu) / sigma

# x1,y1,z1, x2, y2, z2
def create_segments_for_rnn(df, step, time_steps=200):
    X_train = []
    Y_train = []
    hot_test = []
    for i in range(0, df.shape[0] - time_steps, step):
        xyz_data = df[['x-axis', 'y-axis', 'z-axis']][i:i+time_steps]
        X_train.append(np.array(xyz_data))

        label = stats.mode(df['activity'][i: i + time_steps])[0][0]
        hot_test.append(label)
    #怎样找到 one-hot和字符串的对应关系?
    Y_train = np.asarray(pd.get_dummies(hot_test), dtype = np.float32)
    return np.array(X_train), Y_train

# x1, x2, x3, -----, x(timesteps), y1,...z1,...
def create_segments_for_cnn(df, step, time_steps=200):
    X_train = []
    Y_train = []
    hot_test = []
    for i in range(0, df.shape[0] - time_steps, step):
        xs = df['x-axis'].values[i: i + time_steps]
        ys = df['y-axis'].values[i: i + time_steps]
        zs = df['z-axis'].values[i: i + time_steps]
        X_train.append([xs, ys, zs])

        label = stats.mode(df['activity'][i: i + time_steps])[0][0]
        hot_test.append(label)

    reshaped_segments = np.asarray(X_train, dtype= np.float32).reshape(-1, time_steps, 3)
    #怎样找到 one-hot和字符串的对应关系?
    Y_train = np.asarray(pd.get_dummies(hot_test), dtype = np.float32)
    return reshaped_segments, Y_train

def shuffle_data(X, Y):
    np.random.seed(10)
    randomList = np.arange(X.shape[0])
    np.random.shuffle(randomList)
    return X[randomList], Y[randomList]

def split_data(X,Y,rate):
  X_train = X[int(X.shape[0]*rate):]
  Y_train = Y[int(Y.shape[0]*rate):]
  X_val = X[:int(X.shape[0]*rate)]
  Y_val = Y[:int(Y.shape[0]*rate)]
  return X_train, Y_train, X_val, Y_val

网络模型

rnn-lstm

参数的意义:64:lstm隐藏单元,input_length:time sequence length, input_dim: input feature number

return_sequences:多对多 还是多对一

def build_rnn_model(shape, num_classes):
  model = Sequential()
  model.add(LSTM(64, input_length = shape[1], input_dim = shape[2], return_sequences = True ))
  model.add(LSTM(64, return_sequences = False))
  model.add(Dense(num_classes, activation='softmax'))
  print(model.summary())
  model.compile(loss='categorical_crossentropy',
                optimizer='adam', metrics=['accuracy'])
  return model

cnn

参数的意义:30:提取多少个feature, 10: kenerl size, filter, input_shape:filter要扫描的形状

def build_cnn_model(shape, num_classes):
  model = Sequential()
  model.add(Conv1D(30, 10, activation='relu', input_shape=(shape[1], shape[2])))
  model.add(Conv1D(30, 10, activation='relu'))
  model.add(MaxPooling1D(3))
  model.add(Conv1D(48, 10, activation='relu'))
  model.add(Conv1D(48, 10, activation='relu'))
  model.add(GlobalAveragePooling1D())
  model.add(Dropout(0.5))
  model.add(Dense(num_classes, activation='softmax'))
  print(model.summary())
  model.compile(loss='categorical_crossentropy',
                optimizer='adam', metrics=['accuracy'])
  return model

这里理解下什么是1D cnn, 什么是2D cnn?

https://juejin.im/post/5beb7432f265da61524cf27c

[译] 在 Keras 中使用一维卷积神经网络处理时间序列数据

对应的参考代码

https://github.com/ni79ls/har-keras-cnn/blob/master/20180903_Keras_HAR_WISDM_CNN_v1.0_for_medium.py

无论是一维、二维还是三维,卷积神经网络(CNNs)都具有相同的特点和相同的处理方法。关键区别在于输入数据的维数以及特征检测器(或滤波器)如何在数据之间滑动

 模型training

df = read_data("WISDM_ar_v1.1_raw.txt")
df['x-axis'] = feature_normalize(df['x-axis'])
df['y-axis'] = feature_normalize(df['y-axis'])
df['z-axis'] = feature_normalize(df['z-axis'])
df.head()

#_x, _y = create_segments_for_cnn(df, 40, 80)
_x, _y = create_segments_for_rnn(df, 40, 80)
_xs, _ys = shuffle_data(_x, _y)

x_train, y_train, x_test, y_test = split_data(_xs, _ys, 0.1)
model = build_cnn_model(x_train.shape, 6)

callbacks_list = [
    keras.callbacks.ModelCheckpoint(
        filepath='callback_test.h5',
    ),
    #keras.callbacks.tensorboard(log_dir='my_log_dir', histogram_freq=1,)
]

# Hyper-parameters
BATCH_SIZE = 400
EPOCHS = 30

# Enable validation to use ModelCheckpoint and EarlyStopping callbacks.
history = model.fit(x_train,
                      y_train,
                      batch_size=BATCH_SIZE,
                      epochs=EPOCHS,
                      validation_split=0.2,
                      #callbacks=callbacks_list,not supported now
                      verbose=1)

数据处理方式的选择

不同数据格式训练出的效果不同
_x, _y = create_segments_for_rnn(df, 40, 80) 较好但生成训练数据慢

_x, _y = create_segments_for_cnn(df, 40, 80) 相对较差,但生成训练数据快

 有关callbacks 和tensorboard

当前tf.keras的版本对callbacks支持的不好,实测只有保存h5文件好用, 要使用keras,callbacks

callbacks_list = [
    keras.callbacks.ModelCheckpoint(
        filepath='lstm_model.h5',
    ),
    keras.callbacks.TensorBoard(
        log_dir='my_log_dir',
        histogram_freq=1,
    )
]

图形化表示训练效果

从网上找到一些方法,但报找不到xxx, 只要用history.history.keys()找到对应的字段就可以了

In [54]: history.history.keys()
Out[54]: ['loss', 'val_accuracy', 'val_loss', 'accuracy']

print("\n--- Learning curve of model training ---\n")

# summarize history for accuracy and loss
plt.figure(figsize=(6, 4))
plt.plot(history.history['accuracy'], "g--", label="Accuracy of training data")
plt.plot(history.history['val_accuracy'], "g", label="Accuracy of validation data")
plt.plot(history.history['loss'], "r--", label="Loss of training data")
plt.plot(history.history['val_loss'], "r", label="Loss of validation data")
plt.title('Model Accuracy and Loss')
plt.ylabel('Accuracy and Loss')
plt.xlabel('Training Epoch')
plt.ylim(0)
plt.legend()
plt.show()

测试数据上的效果

score = model.evaluate(x_test, y_test, verbose=1)

保存h5格式model文件

model.save('0710_cnn.h5')

h5转换为tflite格式文件

使用tflite_convert (推荐)


usage: tflite_convert [-h] --output_file OUTPUT_FILE
                      (--saved_model_dir SAVED_MODEL_DIR | --keras_model_file KERAS_MODEL_FILE)

tf.lite.TFLiteConverter.from_keras_model(model)

model = keras.models.load_model('path_to_my_model.h5')

converter = tf.lite.TFLiteConverter.from_keras_model(model)
tflite_model = converter.convert()
open("converted_model.tflite", "wb").write(tflite_model)
 

tflite 模型文件在手机上的使用

修改编译 minimal

bazel build --cxxopt='--std=c++11' //tensorflow/lite/examples/minimal:minimal \
--crosstool_top=//external:android/crosstool \
--host_crosstool_top=@bazel_tools//tools/cpp:toolchain \
--cpu=arm64-v8a

tflite::PrintInterpreterState(interpreter.get());

会打印出这个model的结构如:

Interpreter has 41 tensors and 12 nodes

inputs/outputs


Inputs: 1
Outputs: 0

tensor

Tensor   0 Identity             kTfLiteFloat32  kTfLiteArenaRw         24 bytes ( 0.0 MB)  1 6
Tensor   1 conv1d_input         kTfLiteFloat32  kTfLiteArenaRw        960 bytes ( 0.0 MB)  1 80 3
Tensor   2 sequential/conv1d/Relu kTfLiteFloat32  kTfLiteArenaRw       8520 bytes ( 0.0 MB)  1 1 71 30
Tensor   3 sequential/conv1d/conv1d/ExpandDims kTfLiteFloat32  kTfLiteArenaRw        960 bytes ( 0.0 MB)  1 1 80 3
Tensor   4 sequential/conv1d/conv1d/ExpandDims/dim_0 kTfLiteInt32   kTfLiteMmapRo         16 bytes ( 0.0 MB)  1 4
Tensor   5 sequential/conv1d/conv1d/ExpandDims_1 kTfLiteFloat32   kTfLiteMmapRo       3600 bytes ( 0.0 MB)  30 1 10 3
Tensor   6 sequential/conv1d/conv1d_bias kTfLiteFloat32   kTfLiteMmapRo        120 bytes ( 0.0 MB)  30
Tensor   7 sequential/conv1d_1/conv1d/ExpandDims_1 kTfLiteFloat32   kTfLiteMmapRo      36000 bytes ( 0.0 MB)  30 1 10 30
Tensor   8 sequential/conv1d_1/conv1d/Squeeze kTfLiteFloat32  kTfLiteArenaRw       7440 bytes ( 0.0 MB)  1 1 62 30
Tensor   9 sequential/conv1d_1/conv1d_bias kTfLiteFloat32   kTfLiteMmapRo        120 bytes ( 0.0 MB)  30
Tensor  10 sequential/conv1d_2/conv1d/ExpandDims kTfLiteFloat32  kTfLiteArenaRw       2400 bytes ( 0.0 MB)  1 1 20 30
Tensor  11 sequential/conv1d_2/conv1d/ExpandDims/dim_0 kTfLiteInt32   kTfLiteMmapRo         16 bytes ( 0.0 MB)  1 4
Tensor  12 sequential/conv1d_2/conv1d/ExpandDims_1 kTfLiteFloat32   kTfLiteMmapRo      57600 bytes ( 0.1 MB)  48 1 10 30
Tensor  13 sequential/conv1d_2/conv1d/Squeeze kTfLiteFloat32  kTfLiteArenaRw       2112 bytes ( 0.0 MB)  1 1 11 48
Tensor  14 sequential/conv1d_2/conv1d_bias kTfLiteFloat32   kTfLiteMmapRo        192 bytes ( 0.0 MB)  48
Tensor  15 sequential/conv1d_3/Relu kTfLiteFloat32  kTfLiteArenaRw        384 bytes ( 0.0 MB)  1 2 48
Tensor  16 sequential/conv1d_3/conv1d/ExpandDims_1 kTfLiteFloat32   kTfLiteMmapRo      92160 bytes ( 0.1 MB)  48 1 10 48
Tensor  17 sequential/conv1d_3/conv1d/Squeeze kTfLiteFloat32  kTfLiteArenaRw        384 bytes ( 0.0 MB)  1 1 2 48
Tensor  18 sequential/conv1d_3/conv1d/Squeeze_shape kTfLiteInt32   kTfLiteMmapRo         12 bytes ( 0.0 MB)  3
Tensor  19 sequential/conv1d_3/conv1d_bias kTfLiteFloat32   kTfLiteMmapRo        192 bytes ( 0.0 MB)  48
Tensor  20 sequential/dense/BiasAdd kTfLiteFloat32  kTfLiteArenaRw         24 bytes ( 0.0 MB)  1 6
Tensor  21 sequential/dense/MatMul/ReadVariableOp/transpose kTfLiteFloat32   kTfLiteMmapRo       1152 bytes ( 0.0 MB)  6 48
Tensor  22 sequential/dense/MatMul_bias kTfLiteFloat32   kTfLiteMmapRo         24 bytes ( 0.0 MB)  6
Tensor  23 sequential/global_average_pooling1d/Mean kTfLiteFloat32  kTfLiteArenaRw        192 bytes ( 0.0 MB)  1 48
Tensor  24 sequential/global_average_pooling1d/Mean/reduction_indices kTfLiteInt32   kTfLiteMmapRo          4 bytes ( 0.0 MB) 
Tensor  25 sequential/max_pooling1d/ExpandDims kTfLiteFloat32  kTfLiteArenaRw       7440 bytes ( 0.0 MB)  1 62 1 30
Tensor  26 sequential/max_pooling1d/ExpandDims/dim_0 kTfLiteInt32   kTfLiteMmapRo         16 bytes ( 0.0 MB)  1 4
Tensor  27 sequential/max_pooling1d/MaxPool kTfLiteFloat32  kTfLiteArenaRw       2400 bytes ( 0.0 MB)  1 20 1 30
Tensor  28 (null)               kTfLiteInt32  kTfLiteArenaRw         12 bytes ( 0.0 MB)  3
Tensor  29 (null)               kTfLiteInt32  kTfLiteArenaRw          4 bytes ( 0.0 MB)  1
Tensor  30 (null)               kTfLiteFloat32  kTfLiteArenaRw        192 bytes ( 0.0 MB)  48
Tensor  31 (null)               kTfLiteNoType  kTfLiteMemNone          0 bytes ( 0.0 MB)  (null)
Tensor  32 (null)               kTfLiteNoType  kTfLiteMemNone          0 bytes ( 0.0 MB)  (null)
Tensor  33 (null)               kTfLiteFloat32  kTfLiteArenaRw       8520 bytes ( 0.0 MB)  1 1 71 30
Tensor  34 (null)               kTfLiteFloat32 kTfLiteArenaRwPersistent       3600 bytes ( 0.0 MB)  30 30
Tensor  35 (null)               kTfLiteFloat32  kTfLiteArenaRw      74400 bytes ( 0.1 MB)  1 1 62 300
Tensor  36 (null)               kTfLiteFloat32 kTfLiteArenaRwPersistent      36000 bytes ( 0.0 MB)  300 30
Tensor  37 (null)               kTfLiteFloat32  kTfLiteArenaRw      13200 bytes ( 0.0 MB)  1 1 11 300
Tensor  38 (null)               kTfLiteFloat32 kTfLiteArenaRwPersistent      57600 bytes ( 0.1 MB)  300 48
Tensor  39 (null)               kTfLiteFloat32  kTfLiteArenaRw       3840 bytes ( 0.0 MB)  1 1 2 480
Tensor  40 (null)               kTfLiteFloat32 kTfLiteArenaRwPersistent      92160 bytes ( 0.1 MB)  480 48

Node

Node   0 Operator Builtin Code  22
  Inputs: 1 4
  Outputs: 3
Node   1 Operator Builtin Code   3
  Inputs: 3 5 6
  Outputs: 2
Node   2 Operator Builtin Code   3
  Inputs: 2 7 9
  Outputs: 8
Node   3 Operator Builtin Code  22
  Inputs: 8 26
  Outputs: 25
Node   4 Operator Builtin Code  17
  Inputs: 25
  Outputs: 27
Node   5 Operator Builtin Code  22
  Inputs: 27 11
  Outputs: 10
Node   6 Operator Builtin Code   3
  Inputs: 10 12 14
  Outputs: 13
Node   7 Operator Builtin Code   3
  Inputs: 13 16 19
  Outputs: 17
Node   8 Operator Builtin Code  22
  Inputs: 17 18
  Outputs: 15
Node   9 Operator Builtin Code  40
  Inputs: 15 24
  Outputs: 23
Node  10 Operator Builtin Code   9
  Inputs: 23 21 22
  Outputs: 20
Node  11 Operator Builtin Code  25
  Inputs: 20
  Outputs: 0

Tensor的名字和形状

Tensor   0 Identity             kTfLiteFloat32  kTfLiteArenaRw         24 bytes ( 0.0 MB)  1 6
Tensor   1 conv1d_input         kTfLiteFloat32  kTfLiteArenaRw        960 bytes ( 0.0 MB)  1 80 3

怎样赋值input 

如果输入数据的形状如一维矢量,二维矩阵等,会引起tensor data的赋值方式吗?或者说需要根据shape更改输入数据的表示吗?如用二维数组表示矩阵等。其实没有必要,只要数据存储顺序按照二维的格式就行,可以用np.reshape一下即可

  float sensor_test[] = { //[0, 1, 0...]
          -1.02622092e-01,  1.73495474e+00,  1.96860060e+00, -7.71851445e-03,
         -5.41881331e-01,  7.67533877e-01, -3.54595601e-02, -2.29101321e+00,

        -----

}

     // Fill input buffers
    // TODO(user): Insert code to fill input tensors
    Interpreter *interp = interpreter.get();
    TfLiteTensor* input = interp->tensor(interp->inputs()[0]);
    for (int i=0; i < input->bytes/4; i++) {
      //input->data.f[i] = feature_test[i];
      input->data.f[i] = sensor_test[i];//和一维数据时一样,不用多维数组表示数据
      //input->data.f[i] = tensor_test_1[i];
    }

 

TFLite 模型quantilization化

toco需要是源码编译生成的不是自动安装的,另外后面的参数可以从上面tflite的打印log中看出

从结果看并不是所有的float变为uint8型

./toco --input_file='30_cnn.tflite' --input_format=TFLITE --output_format=TFLITE --output_file='quanized_30_cnn.tflite' --inference_type=FLOAT --input_type=FLOAT --input_arrays=conv1d_input --output_arrays=Identity --input_shapes=1,80,3 --post_training_quantize 
2019-07-11 12:06:26.592093: W tensorflow/lite/toco/toco_cmdline_flags.cc:283] --input_type is deprecated. It was an ambiguous flag that set both --input_data_types and --inference_input_type. If you are trying to complement the input file with information about the type of input arrays, use --input_data_type. If you are trying to control the quantization/dequantization of real-numbers input arrays in the output file, use --inference_input_type.
2019-07-11 12:06:26.594579: I tensorflow/lite/toco/graph_transformations/graph_transformations.cc:39] Before Removing unused ops: 12 operators, 28 arrays (0 quantized)
2019-07-11 12:06:26.594937: I tensorflow/lite/toco/graph_transformations/graph_transformations.cc:39] Before general graph transformations: 12 operators, 28 arrays (0 quantized)
2019-07-11 12:06:26.595742: I tensorflow/lite/toco/graph_transformations/graph_transformations.cc:39] After general graph transformations pass 1: 12 operators, 28 arrays (0 quantized)
2019-07-11 12:06:26.596726: I tensorflow/lite/toco/graph_transformations/graph_transformations.cc:39] Before dequantization graph transformations: 12 operators, 28 arrays (0 quantized)
2019-07-11 12:06:26.597362: I tensorflow/lite/toco/allocate_transient_arrays.cc:345] Total transient array allocated size: 17024 bytes, theoretical optimal value: 16064 bytes.
2019-07-11 12:06:26.597556: I tensorflow/lite/toco/toco_tooling.cc:397] Estimated count of arithmetic ops: 0.00166014 billion (note that a multiply-add is counted as 2 ops).
2019-07-11 12:06:26.598574: I tensorflow/lite/toco/tflite/export.cc:569] Quantizing TFLite model after conversion to flatbuffer. dump_graphviz will only output the model before this transformation. To visualize the output graph use lite/tools/optimize.py.
2019-07-11 12:06:26.603058: I tensorflow/lite/tools/optimize/quantize_weights.cc:199] Skipping quantization of tensor sequential/conv1d/conv1d/ExpandDims_1 because it has fewer than 1024 elements (900).
2019-07-11 12:06:26.603184: I tensorflow/lite/tools/optimize/quantize_weights.cc:278] Quantizing tensor sequential/conv1d_1/conv1d/ExpandDims_1 with 9000 elements for hybrid evaluation.
2019-07-11 12:06:26.603352: I tensorflow/lite/tools/optimize/quantize_weights.cc:278] Quantizing tensor sequential/conv1d_2/conv1d/ExpandDims_1 with 14400 elements for hybrid evaluation.
2019-07-11 12:06:26.603476: I tensorflow/lite/tools/optimize/quantize_weights.cc:278] Quantizing tensor sequential/conv1d_3/conv1d/ExpandDims_1 with 23040 elements for hybrid evaluation.
2019-07-11 12:06:26.603670: I tensorflow/lite/tools/optimize/quantize_weights.cc:199] Skipping quantization of tensor sequential/dense/MatMul/ReadVariableOp/transpose because it has fewer than 1024 elements (288). 

LSTM模型h5转换为tflite格式

原本计划使用lstm,且从tflite的operator也看到lstm, 但转换失败

ValueError: Cannot find the Placeholder op that is an input to the ReadVariableOp.

问题对应的源码如下:如果不是Placeholder, 是什么哪?

   165       if map_name_to_node[input_name].op != "Placeholder":
   166         raise ValueError("Cannot find the Placeholder op that is an input "
   167                          "to the ReadVariableOp.")

直接在源码打log没有输出,原因不明。

使用PyCharm 打断点:map_name_to_node[input_name].op 是Switch

看Graph_def中有大量的Switch, 也许是传说中的控制流吧,直接放弃lstm 改用cnn,从效果看也挺好

这篇关于TFLite:使用1维CNN处理序列数据的过程的文章就介绍到这儿,希望我们推荐的文章对编程师们有所帮助!



http://www.chinasem.cn/article/1027316

相关文章

oracle 11g导入\导出(expdp impdp)之导入过程

《oracle11g导入导出(expdpimpdp)之导入过程》导出需使用SEC.DMP格式,无分号;建立expdir目录(E:/exp)并确保存在;导入在cmd下执行,需sys用户权限;若需修... 目录准备文件导入(impdp)1、建立directory2、导入语句 3、更改密码总结上一个环节,我们讲了

Java使用Javassist动态生成HelloWorld类

《Java使用Javassist动态生成HelloWorld类》Javassist是一个非常强大的字节码操作和定义库,它允许开发者在运行时创建新的类或者修改现有的类,本文将简单介绍如何使用Javass... 目录1. Javassist简介2. 环境准备3. 动态生成HelloWorld类3.1 创建CtC

使用Python批量将.ncm格式的音频文件转换为.mp3格式的实战详解

《使用Python批量将.ncm格式的音频文件转换为.mp3格式的实战详解》本文详细介绍了如何使用Python通过ncmdump工具批量将.ncm音频转换为.mp3的步骤,包括安装、配置ffmpeg环... 目录1. 前言2. 安装 ncmdump3. 实现 .ncm 转 .mp34. 执行过程5. 执行结

Python实现批量CSV转Excel的高性能处理方案

《Python实现批量CSV转Excel的高性能处理方案》在日常办公中,我们经常需要将CSV格式的数据转换为Excel文件,本文将介绍一个基于Python的高性能解决方案,感兴趣的小伙伴可以跟随小编一... 目录一、场景需求二、技术方案三、核心代码四、批量处理方案五、性能优化六、使用示例完整代码七、小结一、

Python中 try / except / else / finally 异常处理方法详解

《Python中try/except/else/finally异常处理方法详解》:本文主要介绍Python中try/except/else/finally异常处理方法的相关资料,涵... 目录1. 基本结构2. 各部分的作用tryexceptelsefinally3. 执行流程总结4. 常见用法(1)多个e

Java使用jar命令配置服务器端口的完整指南

《Java使用jar命令配置服务器端口的完整指南》本文将详细介绍如何使用java-jar命令启动应用,并重点讲解如何配置服务器端口,同时提供一个实用的Web工具来简化这一过程,希望对大家有所帮助... 目录1. Java Jar文件简介1.1 什么是Jar文件1.2 创建可执行Jar文件2. 使用java

C#使用Spire.Doc for .NET实现HTML转Word的高效方案

《C#使用Spire.Docfor.NET实现HTML转Word的高效方案》在Web开发中,HTML内容的生成与处理是高频需求,然而,当用户需要将HTML页面或动态生成的HTML字符串转换为Wor... 目录引言一、html转Word的典型场景与挑战二、用 Spire.Doc 实现 HTML 转 Word1

PHP应用中处理限流和API节流的最佳实践

《PHP应用中处理限流和API节流的最佳实践》限流和API节流对于确保Web应用程序的可靠性、安全性和可扩展性至关重要,本文将详细介绍PHP应用中处理限流和API节流的最佳实践,下面就来和小编一起学习... 目录限流的重要性在 php 中实施限流的最佳实践使用集中式存储进行状态管理(如 Redis)采用滑动

Java中的抽象类与abstract 关键字使用详解

《Java中的抽象类与abstract关键字使用详解》:本文主要介绍Java中的抽象类与abstract关键字使用详解,本文通过实例代码给大家介绍的非常详细,感兴趣的朋友跟随小编一起看看吧... 目录一、抽象类的概念二、使用 abstract2.1 修饰类 => 抽象类2.2 修饰方法 => 抽象方法,没有

MyBatis ParameterHandler的具体使用

《MyBatisParameterHandler的具体使用》本文主要介绍了MyBatisParameterHandler的具体使用,文中通过示例代码介绍的非常详细,对大家的学习或者工作具有一定的参... 目录一、概述二、源码1 关键属性2.setParameters3.TypeHandler1.TypeHa