将基于 Theano 的 Keras 模型定义转换为 TensorFlow
Converting Theano-based Keras model definition to TensorFlow
将基于Theano的Keras模型定义转换为TensorFlow时，是否可以更改输入层上input_shape的顺序？
比如下面一层
Convolution2D(32, 3, 3, input_shape=(3, img_width, img_height))
将被替换为
Convolution2D(32, 3, 3, input_shape=(img_width, img_height, 3))
注意：我不想使用 dim_ordering='th'。
来自Francois Chollet的回答：
I think the question means "what input_shape should I pass to my
first layer given that I'm using TensorFlow and that my default
setting for dim_ordering is "tf"". The answer is yep, that's how you
do it, (img_width, img_height, 3).
Important to note that if you want to load saved models that were
trained with Theano with dim_ordering="th", into a model definition
for TF with dim_ordering="tf", you will need to convert the convolution
kernels. Keras has utils for that.
将基于Theano的Keras模型定义转换为TensorFlow时，是否可以更改输入层上input_shape的顺序？
比如下面一层
Convolution2D(32, 3, 3, input_shape=(3, img_width, img_height))
将被替换为
Convolution2D(32, 3, 3, input_shape=(img_width, img_height, 3))
注意：我不想使用 dim_ordering='th'。
来自Francois Chollet的回答：
I think the question means "what
input_shapeshould I pass to my first layer given that I'm using TensorFlow and that my default setting fordim_orderingis"tf"". The answer is yep, that's how you do it,(img_width, img_height, 3).Important to note that if you want to load saved models that were trained with Theano with
dim_ordering="th", into a model definition for TF withdim_ordering="tf", you will need to convert the convolution kernels. Keras has utils for that.