卷积神经网络模型小案例--多分类识别圆形、三角形、四边形

案例模型

Model: "sequential"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d (Conv2D)              (None, 22, 22, 32)        896       
_________________________________________________________________
conv2d_1 (Conv2D)            (None, 20, 20, 64)        18496     
_________________________________________________________________
max_pooling2d (MaxPooling2D) (None, 10, 10, 64)        0         
_________________________________________________________________
flatten (Flatten)            (None, 6400)              0         
_________________________________________________________________
dense (Dense)                (None, 128)               819328    
_________________________________________________________________
dense_1 (Dense)              (None, 3)                 387       
=================================================================
Total params: 839,107
Trainable params: 839,107
Non-trainable params: 0
_________________________________________________________________
None

案例源码

# 调用要使用的包
import numpy as np
from keras.models import Sequential
from keras.layers import Dense, Flatten
from keras.layers.convolutional import Conv2D, MaxPooling2D
from keras.preprocessing.image import ImageDataGenerator, array_to_img, img_to_array, load_img

# 指定随机种子
np.random.seed(3)

# 生成数据
train_data = ImageDataGenerator(rescale=1./255,
                                rotation_range=30,
                                width_shift_range=0.1,
                                height_shift_range=0.05,
                                shear_range=0.6,
                                zoom_range=0.4,
                                horizontal_flip=True,
                                vertical_flip=True,
                                fill_mode='nearest')


# 数据增强，增加数据集
# img = load_img('D:\\Users\\86138\\Desktop\\train\\triangle\\triangle018.png')
# x = img_to_array(img)
# x = x.reshape((1,) + x.shape)
# i = 0
# for batch in train_data.flow(x, batch_size=1, save_to_dir='D:\\Users\\86138\\Desktop\\train\\_circle', save_prefix='t015', save_format='png'):
#     i += 1
#     if i >= 100:
#         print('ok')
#         break

train_generator = train_data.flow_from_directory(
    'D:\\Users\\86138\\Desktop\\train',    # train数据存储位置
    target_size=(24, 24),
    batch_size=3,
    class_mode='categorical')

test_data = ImageDataGenerator(rescale=1./255)
test_generator = test_data.flow_from_directory(
    'D:\\Users\\86138\\Desktop\\test',    # test数据存储位置
    target_size=(24, 24),
    batch_size=3,
    class_mode='categorical')

# 搭建模型
model = Sequential()
model.add(Conv2D(32, kernel_size=(3,3), activation='relu', input_shape=(24,24,3)))
model.add(Conv2D(64, (3,3), activation='relu'))
model.add(MaxPooling2D(pool_size=(2,2)))
model.add(Flatten())
model.add(Dense(128, activation='relu'))
model.add(Dense(3, activation='softmax'))

# # 生成模型结构
# from keras.models import load_model
# model.save('handwriting.h5')
# print(model.summary())

# 设置模型训练过程
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])

# 训练模型
model.fit_generator(train_generator, steps_per_epoch=1000, epochs=100, validation_data=test_generator, validation_steps=5)
# model.fit(trainX, trainY, batch_size=4, epochs=50)

# 评价模型
print("---Evaluate---")
scores = model.evaluate_generator(test_generator, steps=5)
print("%s: %.2f%%" % (model.metrics_names[0], scores[0]*100))
print("%s: %.2f%%" % (model.metrics_names[1], scores[1]*100))

# 使用模型
print("---Predict---")
output = model.predict_generator(test_generator, steps=5)
np.set_printoptions(formatter={'float' : lambda x: "{0:0.3f}".format(x)})
print(test_generator.class_indices)
print(output)

训练模型

Found 3000 images belonging to 3 classes.
Found 15 images belonging to 3 classes.
WARNING:tensorflow:From <ipython-input-1-d830ac30df85>:64: Model.fit_generator (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.
Instructions for updating:
Please use Model.fit, which supports generators.
Epoch 1/100
1000/1000 [==============================] - 22s 22ms/step - loss: 0.7788 - accuracy: 0.6447 - val_loss: 1.7149 - val_accuracy: 0.6667
Epoch 2/100
1000/1000 [==============================] - 11s 11ms/step - loss: 0.4841 - accuracy: 0.8183 - val_loss: 0.5185 - val_accuracy: 0.7333
Epoch 3/100
1000/1000 [==============================] - 11s 11ms/step - loss: 0.3390 - accuracy: 0.8790 - val_loss: 0.1320 - val_accuracy: 0.9333
Epoch 4/100
1000/1000 [==============================] - 12s 12ms/step - loss: 0.2663 - accuracy: 0.9040 - val_loss: 0.0660 - val_accuracy: 1.0000
...
Epoch 97/100
1000/1000 [==============================] - 11s 11ms/step - loss: 0.0249 - accuracy: 0.9943 - val_loss: 1.5895e-08 - val_accuracy: 1.0000
Epoch 98/100
1000/1000 [==============================] - 12s 12ms/step - loss: 0.0198 - accuracy: 0.9937 - val_loss: 7.9473e-09 - val_accuracy: 1.0000
Epoch 99/100
1000/1000 [==============================] - 11s 11ms/step - loss: 0.0246 - accuracy: 0.9907 - val_loss: 0.0000e+00 - val_accuracy: 1.0000
Epoch 100/100
1000/1000 [==============================] - 11s 11ms/step - loss: 0.0161 - accuracy: 0.9957 - val_loss: 2.1537e-06 - val_accuracy: 1.0000

评价模型

---Evaluate---
WARNING:tensorflow:From <ipython-input-1-d830ac30df85>:69: Model.evaluate_generator (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.
Instructions for updating:
Please use Model.evaluate, which supports generators.
loss: 0.00%
accuracy: 100.00%
---Predict---
WARNING:tensorflow:From <ipython-input-1-d830ac30df85>:75: Model.predict_generator (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.
Instructions for updating:
Please use Model.predict, which supports generators.
{'circle': 0, 'rectangle': 1, 'triangle': 2}
[[0.000 1.000 0.000]
 [0.000 1.000 0.000]
 [1.000 0.000 0.000]
 [0.000 0.000 1.000]
 [1.000 0.000 0.000]
 [0.000 0.000 1.000]
 [1.000 0.000 0.000]
 [0.000 0.000 1.000]
 [1.000 0.000 0.000]
 [0.000 0.000 1.000]
 [0.000 1.000 0.000]
 [0.000 0.000 1.000]
 [0.000 1.000 0.000]
 [1.000 0.000 0.000]
 [0.000 1.000 0.000]]

训练好的模型以及数据集链接
链接：https://pan.baidu.com/s/1q--Ovb2SIVsoKGxHCdh0MQ 提取码：0lyj 复制这段内容后打开百度网盘手机App，操作更方便哦