bernard24 / RIS

Implementation of the approach described in the paper "Recurrent Instance Segmentation" https://arxiv.org/abs/1511.08250.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Inference

xjoshramos opened this issue · comments

How to implement inference?

Hi @bernard24 and @xjoshramos , should we be using 'plants_convlstm.model' or 'plants_pre_lstm.model' in line 34 of infer_example.lua (x = model:forward(input))? I'm getting this error when I try to use either:
unknown Torch class <torch.CudaTensor>
The variable 'model' doesn't seem to be defined in this particular inference script I assume so that users can pick whichever model they please.

Hi @bernard24 & @xjoshramos , I have added some missing modules, require 'cunn' and 'cudnn', which has fixed the above mentioned error. However, while I'm able to torch.load() either 'plants_convlstm.model' or 'plants_pre_lstm.model' and able to print their contents (ie, they are not nil) they do not respond to the forward() command. Is there another way I have to read in a model?
I have opened a new issue addressing this.