UFFParser: Reduce operator not supported

Robotics & Edge Computing Jetson & Embedded Systems Jetson TX2
1

Hi,

I’m using Inceptionv3 as my base model and converted my model to uff for TensorRT implementation. I got this error while trying to load the engine:

UFFParser: Parser error: global_max_pooling2d_1/Max: Reduce operator not supported

Does this mean that the maxpool layer is not supported in the current release of TensorRT? Is there a work around for this?

Thanks.

2

Hi,

Reduce is support with TensorRT 4.

If you are using TensorRT on x86 host, please update your TensorRT to 4.0 for the Reduce support.
For Jetson, please wait for our next JetPack release.

Thanks.