keras：形状 (None, 1) 和 (None, 3) 不兼容

Question

我正在参加这个 kaggle 比赛，我必须将这个 X 射线分为 3 类细菌、病毒或正常。
我不明白为什么它让我给出同样的错误。
图像是 rgb，输出形状是 (none,3) 所以我真的不知道形状是 (none,1) 的东西在哪里。

有人可以帮助我吗？

这是我的代码：将 tensorflow 导入为 tf 从 tensorflow 导入 keras 来自 tensorflow.keras 导入图层

TRAIN_DIR = 'D:/tf/archiveBilanciato/chest_xray/train/PNEUMONIA'
TEST_DIR = 'D:/tf/archiveBilanciato/chest_xray/test'
IMG_SIZE = 224  #224 è quella migliore
IMAGE_SIZE = (IMG_SIZE, IMG_SIZE)
BATCH_SIZE = 32
LR = 1e-3

import os

nt = 0
for folder_name in ("bacteria", "normal","virus"):
    folder_path = os.path.join("D:/tf/NeoArchiveBilanciato/chest_xray", folder_name)
    for fname in os.listdir(folder_path):
        fpath = os.path.join(folder_path, fname)
        nt += 1
    print("Totale immagini di questa categoria: %d" % nt)
    nt = 0


train_ds = tf.keras.preprocessing.image_dataset_from_directory(
    "D:/tf/NeoArchiveBilanciato/chest_xray",
    validation_split=0.2,
    subset="training",
    seed=1337,
    color_mode='rgb',
    image_size=IMAGE_SIZE,
    batch_size=BATCH_SIZE,
)
val_ds = tf.keras.preprocessing.image_dataset_from_directory(
    "D:/tf/NeoArchiveBilanciato/chest_xray",
    validation_split=0.2,
    subset="validation",
    seed=1337,
    color_mode='rgb',
    image_size=IMAGE_SIZE,
    batch_size=BATCH_SIZE,
)

import tensorflow as tf
import numpy as np
from tensorflow import keras
from tensorflow.keras.applications import DenseNet121
from keras.layers import GlobalAveragePooling2D,Dense

def pre_model():
    base_model = tf.keras.applications.DenseNet121(
        weights='imagenet', include_top=False, input_shape=(224, 224, 3))

    x = base_model.output
    x = GlobalAveragePooling2D()(x)
    predictions = Dense(14, activation="softmax")(x)
    pre_model = keras.Model(inputs=base_model.input, outputs=predictions)

    return pre_model

base_model = pre_model()
base_model.load_weights("D:/tf/nih_pretrained_chest_model.h5")
print("base_model")
print(base_model.summary())


base_model.trainable = False
mio_classificatore = Dense(3, activation='softmax')(base_model.layers[-2].output)

print("mio_classificatore.get_shape()")
print(mio_classificatore.get_shape())

nuovo_model = keras.Model(inputs=base_model.input, outputs=mio_classificatore)
print("nuovo_model")
print(nuovo_model.summary())

train = train_ds.prefetch(buffer_size=32)
val = val_ds.prefetch(buffer_size=32)

callbacks = [
    keras.callbacks.ModelCheckpoint("save_at_{epoch}.h5"),
]

nuovo_model.compile(optimizer=keras.optimizers.Adam(LR),
              loss=keras.losses.CategoricalCrossentropy(),
              metrics=[keras.metrics.Accuracy()])

nuovo_model.fit(train,batch_size=32, epochs=14,callbacks=callbacks, validation_data=val)

错误发生在 model.fit(...) 我收到此消息的位置：

ValueError: Shapes (None, 1) and (None, 3) are incompatible

Answer 1

与 keras 的 DataImageGenerator 不同，image_dataset_from_directory 默认为整数标签。

如果要使用categorical_crossentropy损失函数，需要在image_dataset_from_directory()中定义label_mode='categorical'得到One-Hot编码标签

参见documentation here。

keras：形状 (None, 1) 和 (None, 3) 不兼容

keras: Shapes (None, 1) and (None, 3) are incompatible

python

shapes

keras