第T6周：好莱坞明星识别-优快云博客

本文链接：https://blog.youkuaiyun.com/weixin_50792991/article/details/144888012

文为「365天深度学习训练营」内部文章
参考本文所写文章，请在文章开头带上「🔗 声明」

1.设置GPU

from tensorflow       import keras
from tensorflow.keras import layers,models
import os, PIL, pathlib
import matplotlib.pyplot as plt
import tensorflow        as tf
import numpy             as np

gpus = tf.config.list_physical_devices("GPU")

if gpus:
    gpu0 = gpus[0]                                        #如果有多个GPU，仅使用第0个GPU
    tf.config.experimental.set_memory_growth(gpu0, True)  #设置GPU显存用量按需使用
    tf.config.set_visible_devices([gpu0],"GPU")
    
gpus

2.导入数据

data_dir = "E:/T3/48-data"

data_dir = pathlib.Path(data_dir)

image_count = len(list(data_dir.glob('*/*.jpg')))

print("图片总数为：",image_count)

图片总数为： 1800

roses = list(data_dir.glob('Jennifer Lawrence/*.jpg'))
PIL.Image.open(str(roses[0]))

batch_size = 32
img_height = 224
img_width = 224

"""
关于image_dataset_from_directory()的详细介绍可以参考文章：https://mtyjkh.blog.youkuaiyun.com/article/details/117018789
"""
train_ds = tf.keras.preprocessing.image_dataset_from_directory(
    data_dir,
    validation_split=0.1,
    subset="training",
    label_mode = "categorical",
    seed=123,
    image_size=(img_height, img_width),
    batch_size=batch_size)

Found 1800 files belonging to 17 classes.
Using 1620 files for training.

"""
关于image_dataset_from_directory()的详细介绍可以参考文章：https://mtyjkh.blog.youkuaiyun.com/article/details/117018789
"""
val_ds = tf.keras.preprocessing.image_dataset_from_directory(
    data_dir,
    validation_split=0.1,
    subset="validation",
    label_mode = "categorical",
    seed=123,
    image_size=(img_height, img_width),
    batch_size=batch_size)

Found 1800 files belonging to 17 classes.
Using 180 files for validation.

class_names = train_ds.class_names
print(class_names)

['Angelina Jolie', 'Brad Pitt', 'Denzel Washington', 'Hugh Jackman', 'Jennifer Lawrence', 'Johnny Depp', 'Kate Winslet', 'Leonardo DiCaprio', 'Megan Fox', 'Natalie Portman', 'Nicole Kidman', 'Robert Downey Jr', 'Sandra Bullock', 'Scarlett Johansson', 'Tom Cruise', 'Tom Hanks', 'Will Smith']

plt.figure(figsize=(20, 10))

for images, labels in train_ds.take(1):
    for i in range(20):
        ax = plt.subplot(5, 10, i + 1)

        plt.imshow(images[i].numpy().astype("uint8"))
        plt.title(class_names[np.argmax(labels[i])])
        
        plt.axis("off")

for image_batch, labels_batch in train_ds:
    print(image_batch.shape)
    print(labels_batch.shape)
    break

(32, 224, 224, 3)
(32, 17)

AUTOTUNE = tf.data.AUTOTUNE

train_ds = train_ds.cache().shuffle(1000).prefetch(buffer_size=AUTOTUNE)
val_ds = val_ds.cache().prefetch(buffer_size=AUTOTUNE)

"""
关于卷积核的计算不懂的可以参考文章：https://blog.youkuaiyun.com/qq_38251616/article/details/114278995

layers.Dropout(0.4) 作用是防止过拟合，提高模型的泛化能力。
关于Dropout层的更多介绍可以参考文章：https://mtyjkh.blog.youkuaiyun.com/article/details/115826689
"""

model = models.Sequential([
    layers.experimental.preprocessing.Rescaling(1./255, input_shape=(img_height, img_width, 3)),
    
    layers.Conv2D(16, (3, 3), activation='relu', input_shape=(img_height, img_width, 3)), # 卷积层1，卷积核3*3  
    layers.AveragePooling2D((2, 2)),               # 池化层1，2*2采样
    layers.Conv2D(32, (3, 3), activation='relu'),  # 卷积层2，卷积核3*3
    layers.AveragePooling2D((2, 2)),               # 池化层2，2*2采样
    layers.Dropout(0.5),  
    layers.Conv2D(64, (3, 3), activation='relu'),  # 卷积层3，卷积核3*3
    layers.AveragePooling2D((2, 2)),     
    layers.Dropout(0.5),  
    layers.Conv2D(128, (3, 3), activation='relu'),  # 卷积层3，卷积核3*3
    layers.Dropout(0.5), 
    
    layers.Flatten(),                       # Flatten层，连接卷积层与全连接层
    layers.Dense(128, activation='relu'),   # 全连接层，特征进一步提取
    layers.Dense(len(class_names))               # 输出层，输出预期结果
])

model.summary()  # 打印网络结构

WARNING:tensorflow:From D:\dl\envs\pytorch_gpu\lib\site-packages\keras\src\backend.py:873: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.

Model: "sequential"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 rescaling (Rescaling)       (None, 224, 224, 3)       0         
                                                                 
 conv2d (Conv2D)             (None, 222, 222, 16)      448       
                                                                 
 average_pooling2d (Average  (None, 111, 111, 16)      0         
 Pooling2D)                                                      
                                                                 
 conv2d_1 (Conv2D)           (None, 109, 109, 32)      4640      
                                                                 
 average_pooling2d_1 (Avera  (None, 54, 54, 32)        0         
 gePooling2D)                                                    
                                                                 
 dropout (Dropout)           (None, 54, 54, 32)        0         
                                                                 
 conv2d_2 (Conv2D)           (None, 52, 52, 64)        18496     
                                                                 
 average_pooling2d_2 (Avera  (None, 26, 26, 64)        0         
 gePooling2D)                                                    
                                                                 
 dropout_1 (Dropout)         (None, 26, 26, 64)        0         
                                                                 
 conv2d_3 (Conv2D)           (None, 24, 24, 128)       73856     
                                                                 
 dropout_2 (Dropout)         (None, 24, 24, 128)       0         
                                                                 
 flatten (Flatten)           (None, 73728)             0         
                                                                 
 dense (Dense)               (None, 128)               9437312   
                                                                 
 dense_1 (Dense)             (None, 17)                2193      
                                                                 
=================================================================
Total params: 9536945 (36.38 MB)
Trainable params: 9536945 (36.38 MB)
Non-trainable params: 0 (0.00 Byte)
_________________________________________________________________

# 设置初始学习率
initial_learning_rate = 1e-4

lr_schedule = tf.keras.optimizers.schedules.ExponentialDecay(
        initial_learning_rate, 
        decay_steps=60,      # 敲黑板！！！这里是指 steps，不是指epochs
        decay_rate=0.96,     # lr经过一次衰减就会变成 decay_rate*lr
        staircase=True)

# 将指数衰减学习率送入优化器
optimizer = tf.keras.optimizers.Adam(learning_rate=lr_schedule)

model.compile(optimizer=optimizer,
              loss=tf.keras.losses.CategoricalCrossentropy(from_logits=True),
              metrics=['accuracy'])

from tensorflow.keras.callbacks import ModelCheckpoint, EarlyStopping

epochs = 100

# 保存最佳模型参数
checkpointer = ModelCheckpoint('best_model.h5',
                                monitor='val_accuracy',
                                verbose=1,
                                save_best_only=True,
                                save_weights_only=True)

# 设置早停
earlystopper = EarlyStopping(monitor='val_accuracy', 
                             min_delta=0.001,
                             patience=20, 
                             verbose=1)

history = model.fit(train_ds,
                    validation_data=val_ds,
                    epochs=epochs,
                    callbacks=[checkpointer, earlystopper])

Epoch 1/100
51/51 [==============================] - ETA: 0s - loss: 2.8020 - accuracy: 0.1068
Epoch 1: val_accuracy improved from -inf to 0.13889, saving model to best_model.h5
51/51 [==============================] - 6s 101ms/step - loss: 2.8020 - accuracy: 0.1068 - val_loss: 2.7654 - val_accuracy: 0.1389
Epoch 2/100
51/51 [==============================] - ETA: 0s - loss: 2.6995 - accuracy: 0.1426
Epoch 2: val_accuracy improved from 0.13889 to 0.17778, saving model to best_model.h5
51/51 [==============================] - 5s 93ms/step - loss: 2.6995 - accuracy: 0.1426 - val_loss: 2.6468 - val_accuracy: 0.1778
Epoch 3/100
51/51 [==============================] - ETA: 0s - loss: 2.5822 - accuracy: 0.1704
Epoch 3: val_accuracy improved from 0.17778 to 0.18889, saving model to best_model.h5
51/51 [==============================] - 5s 92ms/step - loss: 2.5822 - accuracy: 0.1704 - val_loss: 2.5724 - val_accuracy: 0.1889
Epoch 4/100
51/51 [==============================] - ETA: 0s - loss: 2.4968 - accuracy: 0.2080
Epoch 4: val_accuracy did not improve from 0.18889
51/51 [==============================] - 5s 93ms/step - loss: 2.4968 - accuracy: 0.2080 - val_loss: 2.5844 - val_accuracy: 0.1833
Epoch 5/100
51/51 [==============================] - ETA: 0s - loss: 2.4022 - accuracy: 0.2296
Epoch 5: val_accuracy did not improve from 0.18889
51/51 [==============================] - 5s 91ms/step - loss: 2.4022 - accuracy: 0.2296 - val_loss: 2.5602 - val_accuracy: 0.1833
Epoch 6/100
51/51 [==============================] - ETA: 0s - loss: 2.3639 - accuracy: 0.2222
Epoch 6: val_accuracy did not improve from 0.18889
51/51 [==============================] - 5s 92ms/step - loss: 2.3639 - accuracy: 0.2222 - val_loss: 2.4889 - val_accuracy: 0.1722
Epoch 7/100
51/51 [==============================] - ETA: 0s - loss: 2.2985 - accuracy: 0.2543
Epoch 7: val_accuracy improved from 0.18889 to 0.22778, saving model to best_model.h5
51/51 [==============================] - 5s 95ms/step - loss: 2.2985 - accuracy: 0.2543 - val_loss: 2.4399 - val_accuracy: 0.2278
Epoch 8/100
51/51 [==============================] - ETA: 0s - loss: 2.2159 - accuracy: 0.2858
Epoch 8: val_accuracy did not improve from 0.22778
51/51 [==============================] - 5s 92ms/step - loss: 2.2159 - accuracy: 0.2858 - val_loss: 2.5525 - val_accuracy: 0.1889
Epoch 9/100
51/51 [==============================] - ETA: 0s - loss: 2.1468 - accuracy: 0.3154
Epoch 9: val_accuracy did not improve from 0.22778
51/51 [==============================] - 5s 93ms/step - loss: 2.1468 - accuracy: 0.3154 - val_loss: 2.4422 - val_accuracy: 0.2222
Epoch 10/100
51/51 [==============================] - ETA: 0s - loss: 2.0766 - accuracy: 0.3278
Epoch 10: val_accuracy did not improve from 0.22778
51/51 [==============================] - 5s 104ms/step - loss: 2.0766 - accuracy: 0.3278 - val_loss: 2.3985 - val_accuracy: 0.2111
Epoch 11/100
51/51 [==============================] - ETA: 0s - loss: 1.9994 - accuracy: 0.3506
Epoch 11: val_accuracy did not improve from 0.22778
51/51 [==============================] - 5s 104ms/step - loss: 1.9994 - accuracy: 0.3506 - val_loss: 2.4852 - val_accuracy: 0.1833
Epoch 12/100
51/51 [==============================] - ETA: 0s - loss: 1.9674 - accuracy: 0.3710
Epoch 12: val_accuracy did not improve from 0.22778
51/51 [==============================] - 5s 104ms/step - loss: 1.9674 - accuracy: 0.3710 - val_loss: 2.4769 - val_accuracy: 0.2222
Epoch 13/100
51/51 [==============================] - ETA: 0s - loss: 1.8588 - accuracy: 0.3975
Epoch 13: val_accuracy did not improve from 0.22778
51/51 [==============================] - 5s 102ms/step - loss: 1.8588 - accuracy: 0.3975 - val_loss: 2.4093 - val_accuracy: 0.2222
Epoch 14/100
51/51 [==============================] - ETA: 0s - loss: 1.8185 - accuracy: 0.4117
Epoch 14: val_accuracy improved from 0.22778 to 0.24444, saving model to best_model.h5
51/51 [==============================] - 5s 101ms/step - loss: 1.8185 - accuracy: 0.4117 - val_loss: 2.3769 - val_accuracy: 0.2444
Epoch 15/100
51/51 [==============================] - ETA: 0s - loss: 1.7363 - accuracy: 0.4574
Epoch 15: val_accuracy did not improve from 0.24444
51/51 [==============================] - 5s 101ms/step - loss: 1.7363 - accuracy: 0.4574 - val_loss: 2.3971 - val_accuracy: 0.2333
Epoch 16/100
51/51 [==============================] - ETA: 0s - loss: 1.6782 - accuracy: 0.4654
Epoch 16: val_accuracy did not improve from 0.24444
51/51 [==============================] - 5s 102ms/step - loss: 1.6782 - accuracy: 0.4654 - val_loss: 2.4277 - val_accuracy: 0.2444
Epoch 17/100
51/51 [==============================] - ETA: 0s - loss: 1.5884 - accuracy: 0.4895
Epoch 17: val_accuracy improved from 0.24444 to 0.26111, saving model to best_model.h5
51/51 [==============================] - 5s 103ms/step - loss: 1.5884 - accuracy: 0.4895 - val_loss: 2.4789 - val_accuracy: 0.2611
Epoch 18/100
51/51 [==============================] - ETA: 0s - loss: 1.5270 - accuracy: 0.5123
Epoch 18: val_accuracy did not improve from 0.26111
51/51 [==============================] - 5s 99ms/step - loss: 1.5270 - accuracy: 0.5123 - val_loss: 2.3671 - val_accuracy: 0.2444
Epoch 19/100
51/51 [==============================] - ETA: 0s - loss: 1.4880 - accuracy: 0.5259
Epoch 19: val_accuracy did not improve from 0.26111
51/51 [==============================] - 5s 99ms/step - loss: 1.4880 - accuracy: 0.5259 - val_loss: 2.4830 - val_accuracy: 0.2389
Epoch 20/100
51/51 [==============================] - ETA: 0s - loss: 1.4282 - accuracy: 0.5389
Epoch 20: val_accuracy did not improve from 0.26111
51/51 [==============================] - 5s 99ms/step - loss: 1.4282 - accuracy: 0.5389 - val_loss: 2.4671 - val_accuracy: 0.2167
Epoch 21/100
51/51 [==============================] - ETA: 0s - loss: 1.3409 - accuracy: 0.5667
Epoch 21: val_accuracy did not improve from 0.26111
51/51 [==============================] - 5s 102ms/step - loss: 1.3409 - accuracy: 0.5667 - val_loss: 2.4547 - val_accuracy: 0.2222
Epoch 22/100
51/51 [==============================] - ETA: 0s - loss: 1.3315 - accuracy: 0.5710
Epoch 22: val_accuracy did not improve from 0.26111
51/51 [==============================] - 5s 100ms/step - loss: 1.3315 - accuracy: 0.5710 - val_loss: 2.5604 - val_accuracy: 0.2444
Epoch 23/100
51/51 [==============================] - ETA: 0s - loss: 1.2629 - accuracy: 0.5951
Epoch 23: val_accuracy did not improve from 0.26111
51/51 [==============================] - 5s 99ms/step - loss: 1.2629 - accuracy: 0.5951 - val_loss: 2.4737 - val_accuracy: 0.2278
Epoch 24/100
51/51 [==============================] - ETA: 0s - loss: 1.2116 - accuracy: 0.6148
Epoch 24: val_accuracy improved from 0.26111 to 0.27778, saving model to best_model.h5
51/51 [==============================] - 5s 100ms/step - loss: 1.2116 - accuracy: 0.6148 - val_loss: 2.4719 - val_accuracy: 0.2778
Epoch 25/100
51/51 [==============================] - ETA: 0s - loss: 1.1305 - accuracy: 0.6451
Epoch 25: val_accuracy did not improve from 0.27778
51/51 [==============================] - 5s 99ms/step - loss: 1.1305 - accuracy: 0.6451 - val_loss: 2.4864 - val_accuracy: 0.2611
Epoch 26/100
51/51 [==============================] - ETA: 0s - loss: 1.0830 - accuracy: 0.6586
Epoch 26: val_accuracy did not improve from 0.27778
51/51 [==============================] - 5s 101ms/step - loss: 1.0830 - accuracy: 0.6586 - val_loss: 2.4766 - val_accuracy: 0.2556
Epoch 27/100
51/51 [==============================] - ETA: 0s - loss: 1.0328 - accuracy: 0.6704
Epoch 27: val_accuracy did not improve from 0.27778
51/51 [==============================] - 5s 101ms/step - loss: 1.0328 - accuracy: 0.6704 - val_loss: 2.6766 - val_accuracy: 0.2722
Epoch 28/100
51/51 [==============================] - ETA: 0s - loss: 0.9848 - accuracy: 0.6827
Epoch 28: val_accuracy did not improve from 0.27778
51/51 [==============================] - 5s 100ms/step - loss: 0.9848 - accuracy: 0.6827 - val_loss: 2.5959 - val_accuracy: 0.2778
Epoch 29/100
51/51 [==============================] - ETA: 0s - loss: 0.9363 - accuracy: 0.7086
Epoch 29: val_accuracy did not improve from 0.27778
51/51 [==============================] - 5s 99ms/step - loss: 0.9363 - accuracy: 0.7086 - val_loss: 2.6760 - val_accuracy: 0.2444
Epoch 30/100
51/51 [==============================] - ETA: 0s - loss: 0.9077 - accuracy: 0.7173
Epoch 30: val_accuracy did not improve from 0.27778
51/51 [==============================] - 5s 99ms/step - loss: 0.9077 - accuracy: 0.7173 - val_loss: 2.5857 - val_accuracy: 0.2667
Epoch 31/100
51/51 [==============================] - ETA: 0s - loss: 0.8526 - accuracy: 0.7352
Epoch 31: val_accuracy improved from 0.27778 to 0.28333, saving model to best_model.h5
51/51 [==============================] - 5s 103ms/step - loss: 0.8526 - accuracy: 0.7352 - val_loss: 2.6705 - val_accuracy: 0.2833
Epoch 32/100
51/51 [==============================] - ETA: 0s - loss: 0.8245 - accuracy: 0.7463
Epoch 32: val_accuracy did not improve from 0.28333
51/51 [==============================] - 5s 101ms/step - loss: 0.8245 - accuracy: 0.7463 - val_loss: 2.7332 - val_accuracy: 0.2667
Epoch 33/100
51/51 [==============================] - ETA: 0s - loss: 0.7920 - accuracy: 0.7562
Epoch 33: val_accuracy did not improve from 0.28333
51/51 [==============================] - 5s 99ms/step - loss: 0.7920 - accuracy: 0.7562 - val_loss: 2.7227 - val_accuracy: 0.2778
Epoch 34/100
51/51 [==============================] - ETA: 0s - loss: 0.7372 - accuracy: 0.7716
Epoch 34: val_accuracy did not improve from 0.28333
51/51 [==============================] - 5s 99ms/step - loss: 0.7372 - accuracy: 0.7716 - val_loss: 2.7903 - val_accuracy: 0.2667
Epoch 35/100
51/51 [==============================] - ETA: 0s - loss: 0.7194 - accuracy: 0.7698
Epoch 35: val_accuracy did not improve from 0.28333
51/51 [==============================] - 5s 99ms/step - loss: 0.7194 - accuracy: 0.7698 - val_loss: 2.7432 - val_accuracy: 0.2833
Epoch 36/100
51/51 [==============================] - ETA: 0s - loss: 0.6930 - accuracy: 0.7772
Epoch 36: val_accuracy did not improve from 0.28333
51/51 [==============================] - 5s 101ms/step - loss: 0.6930 - accuracy: 0.7772 - val_loss: 2.8837 - val_accuracy: 0.2500
Epoch 37/100
51/51 [==============================] - ETA: 0s - loss: 0.6566 - accuracy: 0.7938
Epoch 37: val_accuracy did not improve from 0.28333
51/51 [==============================] - 5s 100ms/step - loss: 0.6566 - accuracy: 0.7938 - val_loss: 2.8333 - val_accuracy: 0.2833
Epoch 38/100
51/51 [==============================] - ETA: 0s - loss: 0.6490 - accuracy: 0.8012
Epoch 38: val_accuracy did not improve from 0.28333
51/51 [==============================] - 5s 99ms/step - loss: 0.6490 - accuracy: 0.8012 - val_loss: 2.8464 - val_accuracy: 0.2778
Epoch 39/100
51/51 [==============================] - ETA: 0s - loss: 0.6268 - accuracy: 0.8130
Epoch 39: val_accuracy did not improve from 0.28333
51/51 [==============================] - 5s 100ms/step - loss: 0.6268 - accuracy: 0.8130 - val_loss: 2.8161 - val_accuracy: 0.2833
Epoch 40/100
51/51 [==============================] - ETA: 0s - loss: 0.5827 - accuracy: 0.8111
Epoch 40: val_accuracy improved from 0.28333 to 0.29444, saving model to best_model.h5
51/51 [==============================] - 5s 102ms/step - loss: 0.5827 - accuracy: 0.8111 - val_loss: 2.8172 - val_accuracy: 0.2944
Epoch 41/100
51/51 [==============================] - ETA: 0s - loss: 0.5758 - accuracy: 0.8210
Epoch 41: val_accuracy improved from 0.29444 to 0.31667, saving model to best_model.h5
51/51 [==============================] - 5s 103ms/step - loss: 0.5758 - accuracy: 0.8210 - val_loss: 2.8566 - val_accuracy: 0.3167
Epoch 42/100
51/51 [==============================] - ETA: 0s - loss: 0.5230 - accuracy: 0.8469
Epoch 42: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 102ms/step - loss: 0.5230 - accuracy: 0.8469 - val_loss: 2.9240 - val_accuracy: 0.3000
Epoch 43/100
51/51 [==============================] - ETA: 0s - loss: 0.4961 - accuracy: 0.8494
Epoch 43: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 100ms/step - loss: 0.4961 - accuracy: 0.8494 - val_loss: 2.9399 - val_accuracy: 0.3000
Epoch 44/100
51/51 [==============================] - ETA: 0s - loss: 0.5207 - accuracy: 0.8420
Epoch 44: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 100ms/step - loss: 0.5207 - accuracy: 0.8420 - val_loss: 2.9821 - val_accuracy: 0.3056
Epoch 45/100
51/51 [==============================] - ETA: 0s - loss: 0.4713 - accuracy: 0.8537
Epoch 45: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 100ms/step - loss: 0.4713 - accuracy: 0.8537 - val_loss: 3.0480 - val_accuracy: 0.3000
Epoch 46/100
51/51 [==============================] - ETA: 0s - loss: 0.4518 - accuracy: 0.8630
Epoch 46: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 101ms/step - loss: 0.4518 - accuracy: 0.8630 - val_loss: 3.0340 - val_accuracy: 0.3056
Epoch 47/100
51/51 [==============================] - ETA: 0s - loss: 0.4353 - accuracy: 0.8759
Epoch 47: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 101ms/step - loss: 0.4353 - accuracy: 0.8759 - val_loss: 3.0947 - val_accuracy: 0.2889
Epoch 48/100
51/51 [==============================] - ETA: 0s - loss: 0.4439 - accuracy: 0.8512
Epoch 48: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 100ms/step - loss: 0.4439 - accuracy: 0.8512 - val_loss: 3.0633 - val_accuracy: 0.2944
Epoch 49/100
51/51 [==============================] - ETA: 0s - loss: 0.4208 - accuracy: 0.8722
Epoch 49: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 100ms/step - loss: 0.4208 - accuracy: 0.8722 - val_loss: 3.0615 - val_accuracy: 0.3000
Epoch 50/100
51/51 [==============================] - ETA: 0s - loss: 0.4005 - accuracy: 0.8864
Epoch 50: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 100ms/step - loss: 0.4005 - accuracy: 0.8864 - val_loss: 3.1585 - val_accuracy: 0.3000
Epoch 51/100
51/51 [==============================] - ETA: 0s - loss: 0.3905 - accuracy: 0.8802
Epoch 51: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 102ms/step - loss: 0.3905 - accuracy: 0.8802 - val_loss: 3.1790 - val_accuracy: 0.2722
Epoch 52/100
51/51 [==============================] - ETA: 0s - loss: 0.3927 - accuracy: 0.8809
Epoch 52: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 100ms/step - loss: 0.3927 - accuracy: 0.8809 - val_loss: 3.1203 - val_accuracy: 0.2944
Epoch 53/100
51/51 [==============================] - ETA: 0s - loss: 0.4025 - accuracy: 0.8722
Epoch 53: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 99ms/step - loss: 0.4025 - accuracy: 0.8722 - val_loss: 3.1396 - val_accuracy: 0.3111
Epoch 54/100
51/51 [==============================] - ETA: 0s - loss: 0.3757 - accuracy: 0.8852
Epoch 54: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 99ms/step - loss: 0.3757 - accuracy: 0.8852 - val_loss: 3.1843 - val_accuracy: 0.3000
Epoch 55/100
51/51 [==============================] - ETA: 0s - loss: 0.3513 - accuracy: 0.8944
Epoch 55: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 101ms/step - loss: 0.3513 - accuracy: 0.8944 - val_loss: 3.1917 - val_accuracy: 0.3111
Epoch 56/100
51/51 [==============================] - ETA: 0s - loss: 0.3424 - accuracy: 0.8907
Epoch 56: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 102ms/step - loss: 0.3424 - accuracy: 0.8907 - val_loss: 3.2322 - val_accuracy: 0.2889
Epoch 57/100
51/51 [==============================] - ETA: 0s - loss: 0.3271 - accuracy: 0.9019
Epoch 57: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 100ms/step - loss: 0.3271 - accuracy: 0.9019 - val_loss: 3.1764 - val_accuracy: 0.3000
Epoch 58/100
51/51 [==============================] - ETA: 0s - loss: 0.3199 - accuracy: 0.8963
Epoch 58: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 101ms/step - loss: 0.3199 - accuracy: 0.8963 - val_loss: 3.2644 - val_accuracy: 0.3111
Epoch 59/100
51/51 [==============================] - ETA: 0s - loss: 0.2990 - accuracy: 0.9117
Epoch 59: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 99ms/step - loss: 0.2990 - accuracy: 0.9117 - val_loss: 3.2755 - val_accuracy: 0.3000
Epoch 60/100
51/51 [==============================] - ETA: 0s - loss: 0.3111 - accuracy: 0.9093
Epoch 60: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 101ms/step - loss: 0.3111 - accuracy: 0.9093 - val_loss: 3.3149 - val_accuracy: 0.3111
Epoch 61/100
51/51 [==============================] - ETA: 0s - loss: 0.3015 - accuracy: 0.9025
Epoch 61: val_accuracy did not improve from 0.31667
51/51 [==============================] - 5s 102ms/step - loss: 0.3015 - accuracy: 0.9025 - val_loss: 3.3743 - val_accuracy: 0.3167
Epoch 61: early stopping

acc = history.history['accuracy']
val_acc = history.history['val_accuracy']

loss = history.history['loss']
val_loss = history.history['val_loss']

epochs_range = range(len(loss))

plt.figure(figsize=(12, 4))
plt.subplot(1, 2, 1)
plt.plot(epochs_range, acc, label='Training Accuracy')
plt.plot(epochs_range, val_acc, label='Validation Accuracy')
plt.legend(loc='lower right')
plt.title('Training and Validation Accuracy')

plt.subplot(1, 2, 2)
plt.plot(epochs_range, loss, label='Training Loss')
plt.plot(epochs_range, val_loss, label='Validation Loss')
plt.legend(loc='upper right')
plt.title('Training and Validation Loss')
plt.show()

# 加载效果最好的模型权重
model.load_weights('best_model.h5')

from PIL import Image
import numpy as np

img = Image.open("E:/T3/48-data/Jennifer Lawrence/003_963a3627.jpg")  #这里选择你需要预测的图片
image = tf.image.resize(img, [img_height, img_width])

img_array = tf.expand_dims(image, 0) 

predictions = model.predict(img_array) # 这里选用你已经训练好的模型
print("预测结果为：",class_names[np.argmax(predictions)])

1/1 [==============================] - 0s 66ms/step
预测结果为： Jennifer Lawrence

心得：调节期间多种方法包括改变损失函数权重等方法均无明显结果考虑网络架构更换或数据增强下一步会继续调整优化学习率