I replace the convolution layer code in example/usr/local/lib/python2.7/dist-packages/tensorrt/examples/tf_to_trt/lenet5.py with slim interface. It can convert the model into UFF format. But the Release Note says ‘the TensorFlow to TensorRT model export does notwork with network models specified using the TensorFlow Slim interface’.
Could I build tensorflow model with slim and then convert it into UFF format?
Hi,
Slim is non-supported.
Is it possible to transfer slim parameter into general tensorflow operation?
Thanks.