层模型需要 1 个输入，但它接收到 2 个输入张量答案

【问题标题】：Layer model expects 1 input(s), but it received 2 input tensors层模型需要 1 个输入，但它接收到 2 个输入张量
【发布时间】：2021-10-08 10:46:36
【问题描述】：

我正在尝试运行以下简单代码。

图像生成器返回两个图像（因此，标签也是图像）。

import numpy as np
import cv2
from tensorflow.keras.models import Model
from tensorflow.keras.layers import Conv2D, Input

def load_image(file):
    image = cv2.imread(file, cv2.IMREAD_UNCHANGED)
    return image

file = './B2.jpeg'

def image_generator(image):
    i = 0
    while True:
        X = image
        y = image
        
        X = np.expand_dims(X, axis=0)
        y = np.expand_dims(y, axis=0)
        
        i = i + 1
        yield [X, y]

inputs = Input(shape=(None, None, 3))
x = Conv2D(filters=3,
           kernel_size=3,
           padding='same',
           activation='relu',
           strides=1)(inputs)

model = Model(inputs=inputs, outputs=x)
model.compile(loss='mae',
              optimizer='adam')


image = load_image(file)
model.fit(image_generator(image), epochs=1)

它给了我：

ValueError: Layer model expects 1 input(s), but it received 2 input tensors. Inputs received: [<tf.Tensor 'IteratorGetNext:0' shape=(None, None, None, None) dtype=uint8>, <tf.Tensor 'IteratorGetNext:1' shape=(None, None, None, None) dtype=uint8>]

I am using tensorflow 2.4.1 and keras 2.4.0

我正在使用这张图片

【问题讨论】：

标签： keras deep-learning tensorflow2.0

【解决方案1】：

在你的情况下：

image_generator(image)

返回<generator object image_generator>。要访问资源，请使用 next(generator_object)。

测试一下，为我工作：

import numpy as np
import cv2
from tensorflow.keras.models import Model
from tensorflow.keras.layers import Conv2D, Input

def load_image(file):
    image = cv2.imread(file, cv2.IMREAD_UNCHANGED)
    return image

file = './image.jpg'

def image_generator(image):
    while True:
        X = image
        y = image
        
        X = np.expand_dims(X, axis=0)
        y = np.expand_dims(y, axis=0)
        
        return (X, y)

inputs = Input(shape=(None, None, 3))
x = Conv2D(filters=3,
           kernel_size=3,
           padding='same',
           activation='relu',
           strides=1)(inputs)

model = Model(inputs=inputs, outputs=x)
model.compile(loss='mae',
              optimizer='adam')


image = load_image(file)
testx,testy = image_generator(image)
model.fit(testx,testy ,epochs=1)

【讨论】：