#-----AI2.py import cv2 import random import numpy as np import matplotlib.pyplot as plt from tensorflow.keras.preprocessing import image from tensorflow.keras.optimizers import RMSprop from tensorflow import keras import tensorflow as tf import os import math import shutil from tensorflow.keras.models import Sequential from tensorflow.keras.layers import Dense from tensorflow.keras.layers import Flatten from tensorflow.keras.layers import Conv2D from tensorflow.keras.layers import MaxPooling2D from tensorflow.keras.layers import Dropout from tensorflow.keras.utils import to_categorical from sklearn.model_selection import train_test_split # 指定亂數種子 seed = 7 np.random.seed(seed) print('1st cell done...') # read image image_size=128 def load_images_from_folder(folder,nfile): images = [] for filename in os.listdir(folder): img = cv2.imread(os.path.join(folder,filename), cv2.IMREAD_GRAYSCALE) if img is not None: img = cv2.resize(img,(image_size,image_size))/255 images.append(img) return np.array(images[:nfile]) # set no. of images for each lable path='PNG'; ncat=9; nfile=110; mod1='model_NUM9C110.h5' ntest=ncat*5; pathT=path+'/T/' LAB=['1','2','3','4','5','6','7','8','9'] print('load path=',path) train = load_images_from_folder(path+'/N1/',nfile) print('type(train)=',type(train),train.shape) Y_train = np.zeros(train.shape[0]) X_train = train train = load_images_from_folder(path+'/N2/',nfile) Y_train = np.append(Y_train, np.ones(train.shape[0])*1) X_train = np.append(X_train, train, axis=0) if(ncat>2): train = load_images_from_folder(path+'/N3/',nfile) Y_train = np.append(Y_train, np.ones(train.shape[0])*2) X_train = np.append(X_train, train, axis=0) if(ncat>3): train = load_images_from_folder(path+'/N4/',nfile) Y_train = np.append(Y_train, np.ones(train.shape[0])*3) X_train = np.append(X_train, train, axis=0) if(ncat>4): train = load_images_from_folder(path+'/N5/',nfile) Y_train = np.append(Y_train, np.ones(train.shape[0])*4) X_train = np.append(X_train, train, axis=0) if(ncat>5): train = load_images_from_folder(path+'/N6/',nfile) Y_train = np.append(Y_train, np.ones(train.shape[0])*5) X_train = np.append(X_train, train, axis=0) if(ncat>6): train = load_images_from_folder(path+'/N7/',nfile) Y_train = np.append(Y_train, np.ones(train.shape[0])*6) X_train = np.append(X_train, train, axis=0) if(ncat>7): train = load_images_from_folder(path+'/N8/',nfile) Y_train = np.append(Y_train, np.ones(train.shape[0])*7) X_train = np.append(X_train, train, axis=0) if(ncat>8): train = load_images_from_folder(path+'/N9/',nfile) Y_train = np.append(Y_train, np.ones(train.shape[0])*8) X_train = np.append(X_train, train, axis=0) if(ncat>9): train = load_images_from_folder(path+'/N10/',nfile) Y_train = np.append(Y_train, np.ones(train.shape[0])*9) X_train = np.append(X_train, train, axis=0) print('X_train.shape_5=',X_train.shape); print('y_train.shape_5=',Y_train.shape) #print(Y_train) # One-hot編碼 Y_train = to_categorical(Y_train) print('Y_train.shape2=',Y_train.shape) #print('Y2_train=',Y_train) #print(Y_train) # 將圖片轉換成 4D 張量 X_train = X_train.reshape(X_train.shape[0], image_size, image_size, 1).astype("float32") X_train, X_test, Y_train, Y_test = train_test_split(X_train, Y_train, test_size=0.1, random_state=42) print('show TEST dat : Fig and Label') for i in range(0,10): img = X_train[i,:,:,0] print(i,img.shape) cv2.imshow('img',img) cv2.waitKey(0) # 定義模型 model = Sequential() model.add(Conv2D(16,(3,3),activation = 'relu', input_shape = (image_size, image_size,1))) model.add(MaxPooling2D(pool_size=(2,2))) model.add(Conv2D(32,(3,3),activation = 'relu')) model.add(MaxPooling2D(pool_size=(2,2))) model.add(Flatten()) model.add(Dense(64,activation='relu')) model.add(Dropout(0.2)) model.add(Dense(32,activation='relu')) model.add(Dense(ncat,activation='softmax')) model.summary() model.compile(loss = 'categorical_crossentropy', optimizer = "adam", metrics = ['accuracy']) print(X_train.shape) print(Y_train.shape) history = model.fit(X_train,Y_train,shuffle=True, validation_split=0.1, batch_size=128, epochs=60) # 顯示訓練和驗證準確度 使用accuracy instead of acc acc = history.history["accuracy"] epochs = range(1, len(acc)+1) val_acc = history.history["val_accuracy"] plt.plot(epochs, acc, "bo-", label="Training Acc") plt.plot(epochs, val_acc, "ro--", label="Validation Acc") plt.title("Training and Validation Accuracy") plt.xlabel("Epochs") plt.ylabel("Accuracy") plt.legend() plt.show() # 顯示訓練和驗證損失 loss = history.history["loss"] epochs = range(1, len(loss)+1) val_loss = history.history["val_loss"] plt.plot(epochs, loss, "bo-", label="Training Loss") plt.plot(epochs, val_loss, "ro--", label="Validation Loss") plt.title("Training and Validation Loss") plt.xlabel("Epochs") plt.ylabel("Loss") plt.legend() plt.show() # 評估模型 print(X_train.shape) print(Y_train.shape) print(X_test.shape) print(Y_test.shape) print("\nTesting ...") loss, accuracy = model.evaluate(X_train, Y_train, verbose=0) print("訓練資料集的準確度 = {:.2f}".format(accuracy)) loss, accuracy = model.evaluate(X_test, Y_test, verbose=0) print("測試資料集的準確度 = {:.2f}".format(accuracy)) model.save(mod1) print('model ',mod1,' is saved...') #------- you split this into a new cell if you want a frequent test run------- model = keras.models.load_model(mod1) print('load ',mod1,' ... done') # read images test = load_images_from_folder(pathT,ntest) print('load is done path=',pathT) X_test = test X_test = X_test.reshape(X_test.shape[0], image_size, image_size, 1).astype("float32") print('X_test:type,len,shape=',type(X_test),len(X_test),X_test.shape) pred = np.argmax(model.predict(X_test),axis=1) print() #print(X_test) print('pred=',pred) for i in range(ntest): img = X_test[i,:,:,0] print(i,'pred[i]=',pred[i],' LAB=',LAB[pred[i]]) cv2.imshow('img',img) cv2.waitKey(0)