Tensorrt can't support Tensorflow layer

AI & Data Science Deep Learning (Training & Inference) TensorRT
1

Dear sir,
I create a NASNet model from the keras.application.NasnetMobile.
But after I froze&transfer it to the uff model. It can’t pass parser by the UFFParser and trow a error like
“[TensorRT] ERROR: UFFParser: Parser error: cropping2d/strided_slice: StridedSlice only support Constant nodes as input for now”
How could I make it to work?

Thanks

2

Hello,

For strided_slice, UFF parser only supports constant input.

input_ needs to be a constant layer rather than variable.

3

Thanks for the feedback. But how can I get the NASnet work on the tensorrt？
Is there any tips to workaround？For example, tips when feeze the net？