TensorFlow（9）（项目）人马图像分类（卷积神经网络）
目錄
基礎(chǔ)理論
1、sigmoid激活函數(shù)
2、聚類&&分類
一、準(zhǔn)備數(shù)據(jù)
1、創(chuàng)建兩個數(shù)據(jù)生成器
2、創(chuàng)建訓(xùn)練數(shù)據(jù)與測試數(shù)據(jù)生成器
?訓(xùn)練數(shù)據(jù)生成器
測試數(shù)據(jù)生成器????????
二、構(gòu)建神經(jīng)網(wǎng)絡(luò)
三、編譯、訓(xùn)練
四、預(yù)測（單圖預(yù)測）
1、待預(yù)測圖像處理
1-1、讀取圖像
1-2、BGR轉(zhuǎn)RGB
1-3、顯示圖像
2、圖像數(shù)據(jù)轉(zhuǎn)換??
3、預(yù)測
總代碼
基礎(chǔ)理論
1、sigmoid激活函數(shù)
由于是二分類，所以最后會用到sigmoid激活函數(shù)（0、1兩值）。
2、聚類&&分類
????????聚類(Clustering)：是指把相似的數(shù)據(jù)劃分到一起，具體劃分的時候并不關(guān)心這一類的標(biāo)簽，目標(biāo)就是把相似的數(shù)據(jù)聚合到一起，聚類是一種無監(jiān)督學(xué)習(xí)(Unsupervised Learning)方法。
（聚類：不按標(biāo)簽把相似數(shù)據(jù)聚合到一起）
????????分類(Classification)：是把不同的數(shù)據(jù)劃分開，其過程是通過訓(xùn)練數(shù)據(jù)集獲得一個分類器，再通過分類器去預(yù)測未知數(shù)據(jù)，分類是一種監(jiān)督學(xué)習(xí)(Supervised Learning)方法。
一、準(zhǔn)備數(shù)據(jù)
數(shù)據(jù)下載地址：
https://storage.googleapis.com/laurencemoroney-blog.appspot.com/horse-or-human.zip \
? ? -O /tmp/horse-or-human.zip?
# 1、準(zhǔn)備數(shù)據(jù)
from tensorflow.keras.preprocessing.image import ImageDataGenerator1、創(chuàng)建兩個數(shù)據(jù)生成器
# 1-1、創(chuàng)建兩個數(shù)據(jù)生成器
train_data_gen = ImageDataGenerator(rescale=1/255)
test_data_gen = ImageDataGenerator(rescale=1/255)2、創(chuàng)建訓(xùn)練數(shù)據(jù)與測試數(shù)據(jù)生成器
?訓(xùn)練數(shù)據(jù)生成器
# 訓(xùn)練數(shù)據(jù)生成器（指向訓(xùn)練數(shù)據(jù)文件夾）
train_generator = train_data_gen.flow_from_directory('D:\\Study\\AI\\OpenCV\\horse-human-data\\horse-or-human',target_size = (150,150), batch_size = 32, class_mode = 'binary'
# 目標(biāo)大小 一批的數(shù)量 二分類
)測試數(shù)據(jù)生成器????????
# 測試數(shù)據(jù)生成器（指向測試數(shù)據(jù)文件夾）
test_generator = test_data_gen.flow_from_directory('D:\\Study\\AI\\OpenCV\\horse-human-data\\validation-horse-or-human',target_size = (150,150), batch_size = 32, class_mode = 'binary'
)二、構(gòu)建神經(jīng)網(wǎng)絡(luò)
# 2、構(gòu)建神經(jīng)網(wǎng)絡(luò)
import tensorflow as tf
from tensorflow.keras.optimizers import RMSprop# 神經(jīng)網(wǎng)絡(luò)模型
model = tf.keras.models.Sequential([# 第一層CNNtf.keras.layers.Conv2D(16, (3,3), activation='relu', input_shape=(150, 150, 3)),tf.keras.layers.MaxPooling2D(2, 2),# 第二層CNNtf.keras.layers.Conv2D(32, (3,3), activation='relu'),tf.keras.layers.MaxPooling2D(2,2),# 第三層CNNtf.keras.layers.Conv2D(64, (3,3), activation='relu'),tf.keras.layers.MaxPooling2D(2,2),# 第四層CNNtf.keras.layers.Conv2D(64, (3,3), activation='relu'),tf.keras.layers.MaxPooling2D(2,2),# 第五層CNNtf.keras.layers.Conv2D(64, (3,3), activation='relu'),tf.keras.layers.MaxPooling2D(2,2),# 輸入層tf.keras.layers.Flatten(),# 隱層tf.keras.layers.Dense(512, activation='relu'),# 輸出層：用sigmoid激活函數(shù)二分類（只有一個神經(jīng)元，結(jié)果只有0和1，分別對應(yīng)人和馬）tf.keras.layers.Dense(1, activation='sigmoid')
])
# 可視化
model.summary()Model: "sequential_3" _________________________________________________________________ Layer (type) Output Shape Param # ================================================================= conv2d_15 (Conv2D) (None, 148, 148, 16) 448 _________________________________________________________________ max_pooling2d_15 (MaxPooling (None, 74, 74, 16) 0 _________________________________________________________________ conv2d_16 (Conv2D) (None, 72, 72, 32) 4640 _________________________________________________________________ max_pooling2d_16 (MaxPooling (None, 36, 36, 32) 0 _________________________________________________________________ conv2d_17 (Conv2D) (None, 34, 34, 64) 18496 _________________________________________________________________ max_pooling2d_17 (MaxPooling (None, 17, 17, 64) 0 _________________________________________________________________ conv2d_18 (Conv2D) (None, 15, 15, 64) 36928 _________________________________________________________________ max_pooling2d_18 (MaxPooling (None, 7, 7, 64) 0 _________________________________________________________________ conv2d_19 (Conv2D) (None, 5, 5, 64) 36928 _________________________________________________________________ max_pooling2d_19 (MaxPooling (None, 2, 2, 64) 0 _________________________________________________________________ flatten_3 (Flatten) (None, 256) 0 _________________________________________________________________ dense_6 (Dense) (None, 512) 131584 _________________________________________________________________ dense_7 (Dense) (None, 1) 513 ================================================================= Total params: 229,537 Trainable params: 229,537 Non-trainable params: 0 _________________________________________________________________
三、編譯、訓(xùn)練
# 3、編譯&&訓(xùn)練
model.compile(loss='binary_crossentropy', optimizer=RMSprop(lr=0.001), metrics=['acc'])
model.fit(train_generator, epochs=5, validation_data=test_generator)
# 訓(xùn)練集 迭代次數(shù) 測試集Epoch 1/10 33/33 [==============================] - 18s 544ms/step - loss: 0.5144 - acc: 0.7420 - val_loss: 0.6703 - val_acc: 0.8438 Epoch 2/10 33/33 [==============================] - 18s 548ms/step - loss: 0.1962 - acc: 0.9299 - val_loss: 0.7632 - val_acc: 0.8594 Epoch 3/10 33/33 [==============================] - 18s 558ms/step - loss: 0.1640 - acc: 0.9445 - val_loss: 0.7594 - val_acc: 0.8594 Epoch 4/10 33/33 [==============================] - 18s 551ms/step - loss: 0.0940 - acc: 0.9640 - val_loss: 1.0639 - val_acc: 0.8398 Epoch 5/10 33/33 [==============================] - 18s 544ms/step - loss: 0.0851 - acc: 0.9747 - val_loss: 1.4272 - val_acc: 0.8633 Epoch 6/10 33/33 [==============================] - 18s 551ms/step - loss: 0.0774 - acc: 0.9698 - val_loss: 1.0516 - val_acc: 0.8516 Epoch 7/10 33/33 [==============================] - 19s 562ms/step - loss: 0.0113 - acc: 0.9981 - val_loss: 2.2403 - val_acc: 0.7734 Epoch 8/10 33/33 [==============================] - 18s 549ms/step - loss: 0.0689 - acc: 0.9786 - val_loss: 1.3702 - val_acc: 0.8672 Epoch 9/10 33/33 [==============================] - 18s 549ms/step - loss: 0.0043 - acc: 1.0000 - val_loss: 2.0496 - val_acc: 0.8594 Epoch 10/10 33/33 [==============================] - 18s 556ms/step - loss: 0.0572 - acc: 0.9873 - val_loss: 1.8541 - val_acc: 0.8633
[12]:
<tensorflow.python.keras.callbacks.History at 0x7fa1b05ddcd0>
四、預(yù)測（單圖預(yù)測）
1、待預(yù)測圖像處理
1-1、讀取圖像
import cv2
import matplotlib.pyplot as plt
import numpy as np# (1) 讀取圖像
# img1:馬 img2:人
img1 = cv2.imread('tensorflow_datasets/validation-horse-or-human/horses/horse1-000.png')
img1 = cv2.resize(img1, (150, 150))
img2 = cv2.imread('tensorflow_datasets/validation-horse-or-human/humans/valhuman01-01.png')
img2 = cv2.resize(img2, (150, 150))1-2、BGR轉(zhuǎn)RGB
BGR轉(zhuǎn)RGB（opencv的色彩空間是BGR，plt色彩空間是RGB）。
# (2) BGR轉(zhuǎn)RGB
# OpenCV中圖片像素按照BGR方式排列；而Matpoltlib中圖片按照RGB方式排序
b,g,r = cv2.split(img1) #分離
img1 = cv2.merge([r,g,b]) #合并
b,g,r = cv2.split(img2) #分離
img2 = cv2.merge([r,g,b]) #合并1-3、顯示圖像
# (3) 顯示圖像
f, ax = plt.subplots(1, 2)
ax[0].imshow(img1)
ax[0].set_title("0")
ax[1].imshow(img2)
ax[1].set_title("1")
plt.show()如果不進行BGR轉(zhuǎn)RGB的操作，就會出現(xiàn)這樣色彩錯亂的情況：
?（來自冥界的人馬）
2、圖像數(shù)據(jù)轉(zhuǎn)換??
這里model.predict()的預(yù)測需要更高維度的圖像數(shù)據(jù)，進行一下轉(zhuǎn)換。
# 圖像轉(zhuǎn)數(shù)據(jù)（用來做預(yù)測）
Img1 = np.expand_dims(img1, axis=0) #轉(zhuǎn)三維數(shù)據(jù)(二維轉(zhuǎn)三維)
Img2 = np.expand_dims(img2, axis=0) #轉(zhuǎn)三維數(shù)據(jù)(二維轉(zhuǎn)三維)img：
Img：
3、預(yù)測
# 4、預(yù)測(分別對兩個圖進行預(yù)測)
for i in range(2):if i==0:# 對圖像1做預(yù)測classes = model.predict(Img1, batch_size=10)print(f'{i}號圖片預(yù)測結(jié)果為：', int(classes[0][0]), '馬')elif i==1:# 對圖像2做預(yù)測classes = model.predict(Img2, batch_size=10)print(f'{i}號圖片預(yù)測結(jié)果為：', int(classes[0][0]), '人')總代碼
# 1、準(zhǔn)備數(shù)據(jù)
from tensorflow.keras.preprocessing.image import ImageDataGenerator# 1-1、創(chuàng)建兩個數(shù)據(jù)生成器
train_data_gen = ImageDataGenerator(rescale=1/255)
test_data_gen = ImageDataGenerator(rescale=1/255)# 1-2、創(chuàng)建訓(xùn)練數(shù)據(jù)與測試數(shù)據(jù)生成器
# 訓(xùn)練數(shù)據(jù)生成器（指向訓(xùn)練數(shù)據(jù)文件夾）
train_generator = train_data_gen.flow_from_directory('tensorflow_datasets/horse-or-human/',target_size = (150,150), batch_size = 32, class_mode = 'binary'
# 目標(biāo)大小 一批的數(shù)量 二分類
)# 測試數(shù)據(jù)生成器（指向測試數(shù)據(jù)文件夾）
test_generator = test_data_gen.flow_from_directory('tensorflow_datasets/validation-horse-or-human/',target_size = (150,150), batch_size = 32, class_mode = 'binary'
)# 2、構(gòu)建神經(jīng)網(wǎng)絡(luò)
import tensorflow as tf
from tensorflow.keras.optimizers import RMSprop# 神經(jīng)網(wǎng)絡(luò)模型
model = tf.keras.models.Sequential([# 第一層CNNtf.keras.layers.Conv2D(16, (3, 3), activation='relu', input_shape=(150, 150, 3)),tf.keras.layers.MaxPooling2D(2, 2),# 第二層CNNtf.keras.layers.Conv2D(32, (3, 3), activation='relu'),tf.keras.layers.MaxPooling2D(2, 2),# 第三層CNNtf.keras.layers.Conv2D(64, (3, 3), activation='relu'),tf.keras.layers.MaxPooling2D(2, 2),# 第四層CNNtf.keras.layers.Conv2D(64, (3, 3), activation='relu'),tf.keras.layers.MaxPooling2D(2, 2),# 第五層CNNtf.keras.layers.Conv2D(64, (3, 3), activation='relu'),tf.keras.layers.MaxPooling2D(2, 2),# 輸入層tf.keras.layers.Flatten(),# 隱層tf.keras.layers.Dense(512, activation='relu'),# 輸出層：用sigmoid激活函數(shù)二分類（只有一個神經(jīng)元，結(jié)果只有0和1，分別對應(yīng)人和馬）tf.keras.layers.Dense(1, activation='sigmoid')
])
# 可視化
model.summary()# 3、編譯&&訓(xùn)練
model.compile(loss='binary_crossentropy', optimizer=RMSprop(lr=0.001), metrics=['acc'])
model.fit(train_generator, epochs=10, validation_data=test_generator)
# 訓(xùn)練集 迭代次數(shù) 測試集# 4、預(yù)測
import cv2
import matplotlib.pyplot as plt
import numpy as np# (1) 讀取圖像
# img1:馬 img2:人
img1 = cv2.imread('tensorflow_datasets/validation-horse-or-human/horses/horse1-000.png')
img1 = cv2.resize(img1, (150, 150))
img2 = cv2.imread('tensorflow_datasets/validation-horse-or-human/humans/valhuman01-01.png')
img2 = cv2.resize(img2, (150, 150))# (2) BGR轉(zhuǎn)RGB
# OpenCV中圖片像素按照BGR方式排列；而Matpoltlib中圖片按照RGB方式排序
b,g,r = cv2.split(img1) #分離
img1 = cv2.merge([r,g,b]) #合并
b,g,r = cv2.split(img2) #分離
img2 = cv2.merge([r,g,b]) #合并# (3) 顯示圖像
f, ax = plt.subplots(1, 2)
ax[0].imshow(img1)
ax[1].imshow(img2)
plt.show()# 圖像轉(zhuǎn)數(shù)據(jù)（用來做預(yù)測）
Img1 = np.expand_dims(img1, axis=0) #轉(zhuǎn)三維數(shù)據(jù)(二維轉(zhuǎn)三維)
Img2 = np.expand_dims(img2, axis=0) #轉(zhuǎn)三維數(shù)據(jù)(二維轉(zhuǎn)三維)# 4、預(yù)測(分別對兩個圖進行預(yù)測)
for i in range(2):if i==0:# 對圖像1做預(yù)測classes = model.predict(Img1, batch_size=10)print(f'{i}號圖片預(yù)測結(jié)果為：', int(classes[0][0]), '馬')elif i==1:# 對圖像2做預(yù)測classes = model.predict(Img2, batch_size=10)print(f'{i}號圖片預(yù)測結(jié)果為：', int(classes[0][0]), '人')總結(jié)
以上是生活随笔為你收集整理的TensorFlow（9）（项目）人马图像分类（卷积神经网络）的全部內(nèi)容，希望文章能夠幫你解決所遇到的問題。
- 上一篇： TensorFlow（8）卷积神经网络实
- 下一篇： python -- plt图表