Negative dimension size caused by subtracting 2 from 1 for 'max_pooling2d_2/MaxPool''MaxPool')

提示错误：

ValueError: Negative dimension size caused by subtracting 2 from 1 for 'max_pooling2d_2/MaxPool' (op: 'MaxPool') with input shapes: [?,1,32,128].

错误代码部分：

model.add(MaxPooling2D(pool_size=(2, 2),strides=(2,2)))

错误原因：

是图片通道的问题。
也就是”channels_last”和”channels_first”数据格式的问题。
input_shape=(3,150, 150)是theano的写法，而tensorflow需要写出：(150,150,3)
https://blog.csdn.net/JohinieLi/article/details/79867898

纠正错误：

**方案1：**在windows下面找到keras的配置文件，C:\Users\zasx-maoshunyi.keras 。

Here is my keras.json
{
"image_data_format" : "channels_first",
"epsilon": 1e-07,
"floatx": "float32",
"backend": "tensorflow"
}

**方案2：**程序开头加入：

    from keras import backend as K  
    K.set_image_dim_ordering('th')  

    from keras import backend as K  
    K.set_image_dim_ordering('tf') 

**方案3：**参见其他大牛的解决方案

model.add(MaxPooling2D(pool_size=(2, 2), dim_ordering="th"))

https://github.com/keras-team/keras/issues/3945