You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

136 lines
6.4 KiB

11 months ago
import tkinter as tk
from PIL import Image, ImageTk
from X1 import *
global Viewcanvas # 定义画布
global Root # 主窗口
global AllModelObj #网络对象
'''
编程16.5编制程序依据AllModelObj和AllModelConn数据结构产生如图16.2的输出界面
目的及编程说明读者通过编程16.5可理解卷积神经网络模型构建的输出界面数据结构及初始值参见编程16.1
'''
def create_instance():
global AllModelObj
global DataSet, Conv, Pool, FullConn, Nonline, Classifier, Error, AjConv, AjFullconn
DataSet = Data_Class("DataSet1", 1, "数据集1", ".", 120, 330)
Conv = Conv_Class("Conv1", 2, "卷积1", ".", 250, 330)
Pool = Pool_Class("Pool1", 3, "最大池化1", ".", 380, 330)
FullConn = FullConn_Class("FullConn1", 4, "全连接1", ".", 510, 330)
Nonline = Nonline_Class("Nonline1", 5, "非线性函数1", ".", 640, 330)
Classifier = Classifier_Class("Classifier1", 6, "分类1", ".", 780, 330)
Error = Error_Class("Error1", 7, "误差计算1", ".", 710, 124)
AjConv = AjConv_Class("AjConv1", 8, "卷积调整1", ".", 250, 70)
AjFullconn = AjFullconn_Class("AjFullconn1", 9, "全连接调整1", ".", 510, 120)
AllModelObj = [DataSet, Conv, Pool, FullConn, Nonline, Classifier, Error, AjConv, AjFullconn]
def connect_class():
global AllModelConn
# 创建连接对象实例
Line1 = ModelConn(1, 1, DataSet.ObjID, Conv.ObjID).output()
Line2 = ModelConn(2, 1, Conv.ObjID, Pool.ObjID).output()
Line3 = ModelConn(3, 1, Pool.ObjID, FullConn.ObjID).output()
Line4 = ModelConn(4, 1, FullConn.ObjID, Nonline.ObjID).output()
Line5 = ModelConn(5, 1, Nonline.ObjID, Classifier.ObjID).output()
Line6 = ModelConn(6, 1, Classifier.ObjID, Error.ObjID).output()
Line7 = ModelConn(7, 2, Error.ObjID, AjFullconn.ObjID).output()
Line8 = ModelConn(8, 2, Error.ObjID, AjConv.ObjID).output()
Line9 = ModelConn(9, 2, AjFullconn.ObjID, FullConn.ObjID).output()
Line10 = ModelConn(10, 2, AjConv.ObjID, Conv.ObjID).output()
# 网络连接对象总表
AllModelConn = [Line1, Line2, Line3, Line4,
Line5, Line6, Line7, Line8,
Line9, Line10]
def element(path):
img = Image.open(path) # 加载图元对应的图片文件
img = img.resize((60, 50)) # 使用resize方法调整图片
img = ImageTk.PhotoImage(img) # 把Image对象转换成PhotoImage对象
Root.img = img # 保存图片的引用,防止被垃圾回收
return img
def window():
global Root
global Viewcanvas
Root = tk.Tk() # 创建一个主窗口
# 设置窗口的大小为1200*750
window_width = 900 # 窗口的宽度
window_height = 550 # 窗口的高度
Root.title("神经网络可视化")
Root.geometry("900x550") # 设置窗口的大小和位置
# 创建一个画布,用于绘制矩形框,设置画布的大小和背景色
Viewcanvas = tk.Canvas(Root, width=window_width, height=window_height, bg="white")
# 将画布添加到主窗口中
Viewcanvas.pack()
# 绘制矩形框,使用不同的颜色和线宽,指定矩形框的左上角和右下角坐标,填充色,边框色和边框宽度
Viewcanvas.create_rectangle(5, 5, 895, 545, fill=None, outline="lightblue", width=2)
def connecting_lines(obj_x, obj_y, text, text_record,image):
Viewcanvas.create_image(obj_x, obj_y, image=image) # 创建图元对象
Viewcanvas.create_text(obj_x + text_record[0], obj_y + text_record[1], text=text, font=("黑体", 14)) # 创建图元对象的标签
def conn_lines(start, end, index):
smooth = [False, True]
width = [2, 4]
if start[0] == end[0]:
Viewcanvas.create_line(start[0], start[1] + 30, end[0] , end[1] - 30, arrow=tk.LAST,
arrowshape=(16, 20, 4), fill='lightblue', smooth=smooth[index], width=width[index])
elif start[1] == end[1]:
Viewcanvas.create_line(start[0] + 30, start[1], end[0] - 30, end[1], arrow=tk.LAST,
arrowshape=(16, 20, 4), fill='lightblue', smooth=smooth[index], width=width[index])
else:
if abs(start[0]-end[0]) > abs(start[1]-end[1]):
# 创建数据线箭头
Viewcanvas.create_line(start[0]-15, start[1], int((start[0] + end[0])/2), end[1], end[0] + 30, end[1], arrow=tk.LAST,
arrowshape=(16, 20, 4), fill='lightblue', smooth=smooth[index], width=width[index])
else:
# 创建数据线箭头
Viewcanvas.create_line(start[0], start[1] - 20, start[0], end[1], end[0] + 30, end[1], arrow=tk.LAST, arrowshape=(16, 20, 4), fill='lightblue', smooth=smooth[index], width=width[index])
def creating_elements():
text_record = [(0, -50), (0, 50), (-80, 0)]
# 遍历AllModelObj列表在窗口左侧创建图元菜单
for obj in AllModelObj:
# 并且要根据需求调整每个对象的位置
obj_x = obj.ObjX # 根据对象的id计算x坐标
obj_y = obj.ObjY # 根据对象的id计算y坐标
Item_Record.append((obj_x, obj_y))
Item_Name.append(obj.ObjID)
# 根据对象的类型,绘制相应的图形
if 'Error' in obj.ObjID:
connecting_lines(obj_x, obj_y, obj.ObjLable, text_record[0], list_image[obj.ObjType - 1])
elif 'Aj' in obj.ObjID:
connecting_lines(obj_x, obj_y, obj.ObjLable, text_record[2], list_image[-1])
else:
connecting_lines(obj_x, obj_y, obj.ObjLable, text_record[1], list_image[obj.ObjType - 1])
def ligature(): # 连接线
# print(Item_Record)
for conn in AllModelConn:
starting = Item_Name.index(conn[2])
# print(starting)
ending = Item_Name.index(conn[3])
if conn[1] == 1:
# print(Item_Record[starting])
conn_lines(Item_Record[starting], Item_Record[ending], 1)
else:
conn_lines(Item_Record[starting], Item_Record[ending], 0)
if __name__ == '__main__':
global AllModelObj
Item_Record = []
Item_Name = []
window()
create_instance()
connect_class()
img_path = ["img/data.png", "img/conv.png", "img/pool.png", "img/full_connect.png", "img/nonlinear.png",
"img/classifier.png", "img/error.png", "img/adjust.png"]
list_image = []
for path in img_path:
list_image.append(element(path))
creating_elements()
ligature()
Root.mainloop()
# print(Item_Record)