python之tensorflow手把手实例讲解斑马线识别实现

 import tensorflow as tf from tensorflow.keras.preprocessing.image import ImageDataGenerator import numpy as np import matplotlib.pyplot as plt import keras 

 train_dir=r'C:\Users\zx\深度学习\Zebra\train' test_dir=r'C:\Users\zx\深度学习\Zebra\test' train_datagen = ImageDataGenerator(rescale=1/255, rotation_range=10,  #旋转 horizontal_flip=True) train_generator = train_datagen.flow_from_directory(train_dir, (50,50), batch_size=1, class_mode='binary', shuffle=False) test_datagen = ImageDataGenerator(rescale=1/255) test_generator = test_datagen.flow_from_directory(test_dir, (50,50), batch_size=1, class_mode='binary', shuffle=False)

 model = tf.keras.models.Sequential([ # 第一层卷积，卷积核为，共16个，输入为150*150*1 tf.keras.layers.Conv2D(16,(3,3),activation='relu',padding='same',input_shape=(50,50,3)), tf.keras.layers.MaxPooling2D((2,2)), # 第二层卷积，卷积核为3*3，共32个， tf.keras.layers.Conv2D(32,(3,3),activation='relu'), tf.keras.layers.MaxPooling2D((2,2)), # 第三层卷积，卷积核为3*3，共64个， tf.keras.layers.Conv2D(64,(3,3),activation='relu'), tf.keras.layers.MaxPooling2D((2,2)), # 第四层卷积，卷积核为3*3，共128个 #     tf.keras.layers.Conv2D(128,(3,3),activation='relu'), #     tf.keras.layers.MaxPooling2D((2,2)), # 数据铺平 tf.keras.layers.Flatten(), tf.keras.layers.Dense(32,activation='relu'), tf.keras.layers.Dense(16,activation='relu'), tf.keras.layers.Dense(2,activation='softmax') ]) print(model.summary()) model.compile(optimize='adam', loss=tf.keras.losses.sparse_categorical_crossentropy, metrics=['acc']) 

 history = model.fit(train_generator， epochs=20, verbose=1) model.save('./Zebra.h5') 

 model.evaluate(test_generator) 

 #可视化 plt.plot(history.history['acc'], label='accuracy') plt.xlabel('Epoch') plt.ylabel('Accuracy') plt.ylim([0.7, 1]) plt.legend(loc='lower right') plt.title('acc') plt.show() 

 pred=model.predict(test_generator) #获取test集的输出 filenames = test_generator.filenames  #获取test数据的文件名 

 for i in range(len(filenames)): if filenames[i][:6]=='others': if np.argmax(pred[i]) != 0: a=filenames[i] plt.figure() print('预测错误的图片：'+a[int(a.find('s')+2):]) print('错误识别为"zebra crossing",正确类型是"others"') print('预测标签为:'+str(np.argmax(pred[i]))+',真实标签为:0') img = plt.imread('Zebra/test/others/'+a[int(a.find('s')+2):]) plt.imshow(img) plt.title(a[int(a.find('s')+2):]) plt.grid(False) break if filenames[i][:6]=='zebra ': if np.argmax(pred[i]) != 1: b= filenames[i] plt.figure() print('预测错误的图片：'+b[int(b.find('g')+2):]) print('错误识别为"others",正确类型是"zebra crossing"') print('预测标签为:'+str(np.argmax(pred[i]))+',真实标签为:1') img = plt.imread('Zebra/test/zebra crossing/'+b[int(b.find('g')+2):]) plt.imshow(img) plt.title(b[int(b.find('g')+2):]) plt.grid(False) break 

 test_generator.labels 

 sum=0 for i in range(len(filenames)): if filenames[i][:6]=='others': if np.argmax(pred[i]) != 0: a=filenames[i] print('预测错误的图片：'+a[int(a.find('s')+2):]+',错误识别为"zebra crossing",正确类型是"others"') sum=sum+1 if filenames[i][:6]=='zebra ': if np.argmax(pred[i]) != 1: b= filenames[i] print('预测错误的图片：'+b[int(b.find('g')+2):]+',错误识别为"others",正确类型是"zebra crossing"') sum=sum+1 print('错误率：'+str(sum/100)+'%') print('正确率：'+str((10000-sum)/100)+'%')