ValueError: Input 0 is incompatible with layer conv2d_1: expected ndim=4, found ndim=2

Question

前提・実現したいこと

以下のコードを実行した結果、エラーが発生しました。
次元に関するエラーなのは分かるのですが、解決方法がわかりません。
ご教授お願いします。

発生している問題・エラーメッセージ

Traceback (most recent call last):
  File "sleepingmodel7.py", line 49, in <module>
    x = layers.Conv2D(32, (3, 3), activation='relu')(bottleneck)
  File "/opt/anaconda3/lib/python3.7/site-packages/keras/engine/base_layer.py", line 446, in __call__
    self.assert_input_compatibility(inputs)
  File "/opt/anaconda3/lib/python3.7/site-packages/keras/engine/base_layer.py", line 342, in assert_input_compatibility
    str(K.ndim(x)))
ValueError: Input 0 is incompatible with layer conv2d_1: expected ndim=4, found ndim=2

該当のソースコード

from keras import layers
from keras import models
from keras import optimizers
from keras.layers import GlobalAveragePooling2D
import numpy as np
import pandas as pd

import os
for dirname, _, filenames in os.walk('/kaggle/input'):
    for filename in filenames:
        print(os.path.join(dirname, filename))

import glob
import matplotlib.pyplot as plt
import seaborn as sns
from PIL import Image
from efficientnet.keras import EfficientNetB7 
from sklearn.model_selection import train_test_split
from keras.utils import to_categorical
from keras.layers import Input, Dense, Reshape
from keras.models import Model
from keras.callbacks import ModelCheckpoint

from keras.preprocessing.image import ImageDataGenerator

train_dir = 'dataset/train'
validation_dir = 'dataset/validation'

# rescaled by 1./255
train_datagen = ImageDataGenerator(
    rescale=1./255#
)

inputs = Input(shape=(100, 100, 3))
efficient_net = EfficientNetB7(weights='noisy-student',include_top=False,input_tensor = inputs,pooling='max')

for layer in efficient_net.layers:
    layer.trainable = True

bottleneck=efficient_net.output
print(bottleneck.shape)
x = layers.Conv2D(32, (3, 3), activation='relu')(bottleneck)
x = layers.MaxPooling2D((2, 2))(x)
x = layers.Conv2D(64, (3, 3), activation='relu')(x)
x = layers.MaxPooling2D((2, 2))(x)
x = layers.Conv2D(128, (3, 3), activation='relu')(x)
x = layers.MaxPooling2D((2, 2))(x)
x = layers.Conv2D(128, (3, 3), activation='relu')(x)
x = layers.MaxPooling2D((2, 2))(x)
x = GlobalAveragePooling2D()(x)
x = layers.Dense(32, activation='relu')(x)
eye_output = layers.Dense(1, activation='sigmoid')(x)

model = Model(inputs=inputs, outputs=eye_output)

model.summary()

model.compile(loss='binary_crossentropy',optimizer=optimizers.RMSprop(lr=1e-4), metrics=['acc'])
test_datagen = ImageDataGenerator(rescale=1./255)

train_generator = train_datagen.flow_from_directory(
        train_dir,
        target_size=(100, 100),
        batch_size=80,
        class_mode='binary')

validation_generator = test_datagen.flow_from_directory(
        validation_dir,
        target_size=(100, 100),
        batch_size=50,
        class_mode='binary')

history = model.fit_generator(
      train_generator,
      steps_per_epoch=30,
      epochs=50,
      validation_data=validation_generator,
      validation_steps=6)

model.save('sleeping1')

from keras.models import load_model
from keras.preprocessing.image import ImageDataGenerator
model = load_model('sleeping1')

test_dir = 'dataset/test'
test_datagen = ImageDataGenerator(rescale=1./255)
test_generator = test_datagen.flow_from_directory(
    test_dir,
    target_size=(100,100),
    batch_size=50,
    class_mode='binary')

test_loss, test_acc = model.evaluate_generator(test_generator, steps=12)
print('test loss:', test_loss)
print('test acc:', test_acc)

Comments

[toast-uz]

再現してみるとエラーは以下のように出ますが、Full shape以降の表示は出ていませんか？

ValueError: Input 0 of layer conv2d is incompatible with the layer: : expected min_ndim=4, found ndim=2. Full shape received: [None, 2560]

単純に読むと、bottleneckの形式がエラーが出たところの入力形式の期待と合っていないことだと思います。bottleneckの形式は期待通りなのでしょうか？

[c7744kumamoto-u]

full shape以降の記述はありませんでした。

conv2dは４次元を期待していて、bottleneckは２次元になっているため期待通りではありません。