ndraw是一个可自由定义网络图、流程图以及简单的神经网络可视化工具

# ndraw [支持作者和技术服务](https://afdian.net/a/huoyo) --- <div > <img src='https://shields.io/badge/version-1.2.7-green.svg'> <img src='https://shields.io/badge/dependencies-tensorflow/metricflow.js/colorpicker.js/dream_msg.js-blue.svg'> <img src='https://shields.io/badge/author-Chang Zhang-dbab09.svg'> <h4>ndraw是一个可自由定义网络图、流程图以及简单的神经网络可视化（目前支持Tensorflow2.0+模型可视化）工具</h4> <h4>A tool to show neural networks and draw custom networks</h4> </div> ## 安装 ### 1、pip安装 ``` pip install ndraw -U ``` ### 2、源码安装 下载源码，打开命令行 ``` python setup.py bdist_egg python setup.py install ``` ## 随手画图 ### 1、启动画图服务 ```python # -*- encoding: utf-8 -*- import ndraw # 打开 http://localhost:43808 根据鼠标即兴画图 ndraw.draw_server(port=43808) # 操作流程看下图 ```  ## 自定义代码画图 ### 1、简单直线流程 ```python # -*- encoding: utf-8 -*- import ndraw graph = ndraw.AutoGraph() graphs = graph.create("开始").to("过程1").to("过程2").to("过程3").to("过程4").to("结束") ndraw.server(graphs,theme=ndraw.BLUE_WHITE) # v1.1.4开始，可以不使用ndraw.server(),AutoGraph自带了server()和render() # graph.create("开始").to("过程1").to("过程2").to("过程3").to("过程4").to("结束").server() ```  ### 2、添加元素信息 > Node ```python # -*- encoding: utf-8 -*- import ndraw from ndraw import Node graph = ndraw.AutoGraph() graphs = graph.create("开始")\ .to(Node("过程1",["1.xxx","2.xxx"]))\ .to(Node("过程2",["1.xxx","2.xxx"]))\ .to("结束") ndraw.server(graphs,theme=ndraw.DEFAULT,flow=ndraw.HORIZONTAL) ```  ### 3、定义单节点样式 > theme ```python # -*- encoding: utf-8 -*- import ndraw from ndraw import Node graph = ndraw.AutoGraph() graphs = graph.create(Node("开始",theme=ndraw.GREEN_WHITE))\ .to(Node("过程1",data=["1.xxx","2.xxx"],theme=ndraw.DEEPGRAY_WHITE))\ .to(Node("过程2",data=["1.xxx","2.xxx"],theme=ndraw.BLUE_WHITE))\ .to(Node("结束",theme=ndraw.GREEN_WHITE)) ndraw.server(graphs) ```  ### 4、分流 > tos() ```python # -*- encoding: utf-8 -*- import ndraw from ndraw import Node graph = ndraw.AutoGraph() graphs = graph.create(Node("开始", theme=ndraw.GREEN_WHITE)) \ .tos([ Node("分流1", data=["1.xxx", "2.xxx"], theme=ndraw.DEEPGRAY_WHITE), Node("分流2", data=["1.xxx", "2.xxx"], theme=ndraw.DEEPGRAY_WHITE), ]) \ .to(Node("汇总", theme=ndraw.GREEN_WHITE))\ .tos([ Node("分流1", data=["1.xxx", "2.xxx"], theme=ndraw.DEEPGRAY_WHITE), Node("分流2", data=["1.xxx", "2.xxx"], theme=ndraw.DEEPGRAY_WHITE), ]) \ .to(Node("结束", theme=ndraw.GREEN_WHITE)) ndraw.server(graphs) ```  ### 5、汇总 > creates() ```python # -*- encoding: utf-8 -*- import ndraw from ndraw import Node graph = ndraw.AutoGraph() graphs = graph.creates([ Node("分流1", data=["1.xxx", "2.xxx"], theme=ndraw.DEEPGRAY_WHITE), Node("分流2", data=["1.xxx", "2.xxx"], theme=ndraw.DEEPGRAY_WHITE), ]) \ .to(Node("结束", theme=ndraw.GREEN_WHITE)) ndraw.server(graphs) ```  ### 6、自定义图形位置 ```python # -*- encoding: utf-8 -*- import ndraw g = ndraw.StableGraph() node1 = g.add_node(ndraw.Node(name='节点1',x=10,y=10)) node2 = g.add_node(ndraw.Node(name='节点2',data =['111111111','22222222'],x=200,y=10)) node3 = g.add_node(ndraw.Node(name='节点3',data =['111111111','22222222'],x=10,y=200)) g.add_link((node1,node2)) g.add_link((node1,node3)) ndraw.server(g,theme=ndraw.DEFAULT) # v1.1.4开始，可以不使用ndraw.server(),StableGraph()和render() # g.server() ``` ### 7、自定义主题 ```python # -*- encoding: utf-8 -*- import ndraw from ndraw import Node class Theme1(ndraw.Theme): ''' 自定义主题 继承ndraw.Theme即可 ''' # 边框颜色 border_color = 'red' # 数据区域颜色 data_color = 'white' # 标题区域颜色 title_color = '#181b16' # 标题字体大小 title_font_size = '17px' # 标题字体颜色 title_font_color = 'white' #其他属性参考 ndraw.Theme graph = ndraw.StableGraph() node1 = graph.add_node(Node("开始",x=10,y=10,theme=Theme1())) ndraw.server(graph) ``` ## TF模型可视化 > 可以参考testTFmodel.py ### 1、pb模型可视化 ``` --pbpath |--variables |--saved_model.pb ``` ```python import ndraw ndraw.server("pbpath") # 打开浏览器访问9999端口即可 ```  ### 2、h5模型可视化 ```python import ndraw ndraw.server("model.h5") # 打开浏览器访问9999端口即可 ``` ### 3、模型对象可视化 ```python import ndraw import tensorflow as tf model = tf.keras.Sequential([ tf.keras.layers.Dense(512, activation='relu'), tf.keras.layers.Dense(256, activation='relu'), tf.keras.layers.Dense(2, activation='softmax') ]) model.build(input_shape=(None, 100)) ndraw.server(model) # 打开浏览器访问9999端口即可 ``` ### 4、生成html文件 ```python import ndraw html = ndraw.render("pb/h5/mode均可",out_file="model.html") # 生成一个model.html文件 ``` ### 5、可视化主题 可视化主题仅仅对于部分节点生效，类似于LSTM，Conv2D已有预置主题 1、DEFAULT ```python import ndraw ndraw.server("model.h5",theme=ndraw.DEFAULT) ```  2、BLACK_WHITE ```python import ndraw ndraw.server("model.h5",theme=ndraw.BLACK_WHITE) ```  3、LIGHTBLACK_WHITE ```python import ndraw ndraw.server("model.h5",theme=ndraw.LIGHTBLACK_WHITE) ```  4、GREEN_WHITE ```python import ndraw ndraw.server("model.h5",theme=ndraw.GREEN_WHITE) ```  6、PURPLE_WHITE ```python import ndraw ndraw.server("model.h5",theme=ndraw.PURPLE_WHITE) ```  ### 6、其他参数 ``` :param host: 服务地址 可自定义 :param port: 服务端口可自定义 :param flow: 布局方式：vertical and horizontal ``` ## 参考图