Fij/Win64. While trying to run demo,I am getting the following error unable to load tensor flow.

[vvelagal]

[HedgehogCode]

Hi @vvelagal,

The info message about not being able to load TensorFlow GPU is because it couldn't load the GPU version of TensorFlow and is falling back to the CPU version (This will still work but is expected to be slower). If you have an capable Nvidia GPU you should install CUDA to make use of it.

The error is not that TensorFlow couldn't be load but that you try to run a model for an image with one channel on an image with 3 channels:

Input axis of type channel should have size 1 but has size 3

Which demo are you trying to execute?

[vvelagal]

hi
I have installed CUDA and I have everything installed with respect to GPU .I am trying to run surface projection.Below are all the paths defined

[HedgehogCode]

@frauzufall can you comment on the setup for GPU support on Windows?

Still, this is not the main issue. The main issue is that the surface projection example was trained for an image with one channel but you are running it on an image with three channels. On which image are you trying to run the model?
Usually, you should use the flywing.tif image from the exemplary image data.