01-CNN案例

kares 创建模型

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
model = Sequential()

model.add(Conv2D(32, kernel_size=(3, 3),
                 activation='relu',
                 input_shape=input_shape))
model.add(Conv2D(64, (3, 3), activation='relu'))
model.add(MaxPooling2D(pool_size=(2, 2)))
model.add(Dropout(0.25))
model.add(Flatten())
model.add(Dense(128, activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(num_classes, activation='softmax'))

model.compile(loss = 'categorical_crossentropy',
              optimizer = keras.optimizers.Adadelta(),
              metrics = ['accuracy'])

print(model.summary())

#训练
history = model.fit(x_train, y_train,
          batch_size=batch_size,
          epochs=epochs,
          verbose=1,
          validation_data=(x_test, y_test))

score = model.evaluate(x_test, y_test, verbose=0)
print('Test loss:', score[0])
print('Test accuracy:', score[1])

#保存模型为文件
model.save('a.h5')

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
x_train shape: (60000, 28, 28, 1)
60000 train samples
10000 test samples
Number of Classes: 10
Model: "sequential_1"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d_1 (Conv2D)            (None, 26, 26, 32)        320       
_________________________________________________________________
conv2d_2 (Conv2D)            (None, 24, 24, 64)        18496     
_________________________________________________________________
max_pooling2d_1 (MaxPooling2 (None, 12, 12, 64)        0         
_________________________________________________________________
dropout_1 (Dropout)          (None, 12, 12, 64)        0         
_________________________________________________________________
flatten_1 (Flatten)          (None, 9216)              0         
_________________________________________________________________
dense_1 (Dense)              (None, 128)               1179776   
_________________________________________________________________
dropout_2 (Dropout)          (None, 128)               0         
_________________________________________________________________
dense_2 (Dense)              (None, 10)                1290      
=================================================================
Total params: 1,199,882
Trainable params: 1,199,882
Non-trainable params: 0
_________________________________________________________________
None
Train on 60000 samples, validate on 10000 samples
Epoch 1/5
60000/60000 [==============================] - 44s 738us/step - loss: 0.2578 - accuracy: 0.9215 - val_loss: 0.0571 - val_accuracy: 0.9808
Epoch 2/5
60000/60000 [==============================] - 44s 736us/step - loss: 0.0878 - accuracy: 0.9737 - val_loss: 0.0411 - val_accuracy: 0.9865
Epoch 3/5
60000/60000 [==============================] - 48s 806us/step - loss: 0.0649 - accuracy: 0.9811 - val_loss: 0.0349 - val_accuracy: 0.9883
Epoch 4/5
60000/60000 [==============================] - 49s 817us/step - loss: 0.0556 - accuracy: 0.9833 - val_loss: 0.0316 - val_accuracy: 0.9891
Epoch 5/5
60000/60000 [==============================] - 49s 823us/step - loss: 0.0474 - accuracy: 0.9855 - val_loss: 0.0332 - val_accuracy: 0.9885
Test loss: 0.0331950055410889
Test accuracy: 0.9884999990463257

kares 加载tf模型

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
import cv2
import numpy as np
from keras.datasets import mnist
from keras.models import load_model

classifier = load_model('/home/zzh/mnist_simple_cnn.h5')

f = np.load('/home/zzh/Downloads/mnist.npz')
x_train, y_train = f['x_train'], f['y_train']
x_test, y_test = f['x_test'], f['y_test']
f.close()

# loads the MNIST dataset
#(x_train, y_train), (x_test, y_test)  = mnist.load_data()

def draw_test(name, pred, input_im):
    BLACK = [0,0,0]
    print(imageL.shape[0])
    expanded_image = cv2.copyMakeBorder(input_im, 0, 0, 0, imageL.shape[0] ,cv2.BORDER_CONSTANT,value=BLACK)
    expanded_image = cv2.cvtColor(expanded_image, cv2.COLOR_GRAY2BGR)
    cv2.putText(expanded_image, str(pred), (152, 70) , cv2.FONT_HERSHEY_COMPLEX_SMALL,4, (0,255,0), 2)
    cv2.imshow(name, expanded_image)

for i in range(0,10):
    rand = np.random.randint(0,len(x_test))
    input_im = x_test[rand]

    imageL = cv2.resize(input_im, None, fx=4, fy=4) 
    input_im = input_im.reshape(1,28,28,1) 

    ## 预测
    res = str(classifier.predict_classes(input_im, 1, verbose = 0)[0])
    draw_test("Prediction", res, imageL) 
    cv2.waitKey(0)

cv2.destroyAllWindows()

image-20191121140200373

识别手写的一组数字

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
import numpy as np
import cv2
from preprocessors import x_cord_contour, makeSquare, resize_to_pixel
from keras.datasets import mnist
from keras.models import load_model

image = cv2.imread('images/image2.png')
gray = cv2.cvtColor(image,cv2.COLOR_BGR2GRAY)
cv2.imshow("image", image)
cv2.waitKey(0)

# 高斯模糊
blurred = cv2.GaussianBlur(gray, (5, 5), 0)
# cv2.imshow("blurred", blurred)
# cv2.waitKey(0)

edged = cv2.Canny(blurred, 30, 150)
# cv2.imshow("edged", edged)
# cv2.waitKey(0)

# 查找边缘
contours, _ = cv2.findContours(edged.copy(), cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)

#从左到右排序坐标
contours = sorted(contours, key = x_cord_contour, reverse = False)

# 存储数据的数组
full_number = []
classifier = load_model('/home/zzh/mnist_simple_cnn.h5')

#循环兴趣点
for c in contours:
    # 矩形边框包过兴趣点
    (x, y, w, h) = cv2.boundingRect(c)    

    if w >= 5 and h >= 25:
        roi = blurred[y:y + h, x:x + w]
        ret, roi = cv2.threshold(roi, 127, 255,cv2.THRESH_BINARY_INV)
        roi = makeSquare(roi)
        roi = resize_to_pixel(28, roi)
        cv2.imshow("ROI", roi)
        roi = roi / 255.0       
        roi = roi.reshape(1,28,28,1) 

        ## 预测
        res = str(classifier.predict_classes(roi, 1, verbose = 0)[0])
        full_number.append(res)
        cv2.rectangle(image, (x, y), (x + w, y + h), (0, 0, 255), 2)
        cv2.putText(image, res, (x , y + 155), cv2.FONT_HERSHEY_COMPLEX, 2, (255, 0, 0), 2)
        cv2.imshow("image", image)
        cv2.waitKey(0) 

cv2.destroyAllWindows()
print ("图片中的数字为: " + ''.join(full_number))

image-20191121140637008

10种物体分类预测

image-20191121150004463

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
import cv2
import numpy as np
from keras.models import load_model
from keras.datasets import cifar10 
from PIL import Image, ImageDraw, ImageFont

img_row, img_height, img_depth = 32,32,3

classifier = load_model('/home/zzh/cifar_simple_cnn.h5')

# Loads the CIFAR dataset
(x_train, y_train), (x_test, y_test) = cifar10.load_data()
color = True 
scale = 4


def paint_chinese_opencv(im,chinese,position,fontsize,color):#opencv输出中文
    img_PIL = Image.fromarray(cv2.cvtColor(im,cv2.COLOR_BGR2RGB))# 图像从OpenCV格式转换成PIL格式
    font = ImageFont.truetype('SimHei.ttf',fontsize,encoding="utf-8")
    #color = (255,0,0) # 字体颜色
    #position = (100,100)# 文字输出位置
    draw = ImageDraw.Draw(img_PIL)
    draw.text(position,chinese,font=font,fill=color)# PIL图片上打印汉字 # 参数1：打印坐标，参数2：文本，参数3：字体颜色，参数4：字体
    img = cv2.cvtColor(np.asarray(img_PIL),cv2.COLOR_RGB2BGR)# PIL图片转cv2 图片
    return img


def draw_test(name, res, input_im, scale, img_row, img_height):
    BLACK = [0,0,0]
    res = int(res)
    if res == 0:
        pred = "灰机"
    if res == 1:
        pred = "汽车"
    if res == 2:
        pred = "鸟"
    if res == 3:
        pred = "猫"
    if res == 4:
        pred = "鹿"
    if res == 5:
        pred = "狗"
    if res == 6:
        pred = "小青蛙"
    if res == 7:
        pred = "马"
    if res == 8:
        pred = "船"
    if res == 9:
        pred = "卡车"

    expanded_image = cv2.copyMakeBorder(input_im, 0, 0, 0, imageL.shape[0] ,cv2.BORDER_CONSTANT,value=BLACK)
    if color == False:
        expanded_image = cv2.cvtColor(expanded_image, cv2.COLOR_GRAY2BGR)
    expanded_image = paint_chinese_opencv(expanded_image,str(pred),(150,40),32,(0,255,0))
    cv2.imshow(name, expanded_image)


for i in range(1,20):
#     rand = np.random.randint(0,len(x_test))
    input_im = x_test[i]
    imageL = cv2.resize(input_im, None, fx=scale, fy=scale, interpolation = cv2.INTER_CUBIC) 
    input_im = input_im.reshape(1,img_row, img_height, img_depth) 

    ## Get Prediction
    res = str(classifier.predict_classes(input_im, 1, verbose = 0)[0])

    draw_test("Prediction", res, imageL, scale, img_row, img_height) 
    cv2.waitKey(0)

cv2.destroyAllWindows()

image-20191121170634201

image-20191121170704953