search

lmb-freiburg / flownet2

FlowNet 2.0: Evolution of Optical Flow Estimation with Deep Networks
https://lmb.informatik.uni-freiburg.de/Publications/2017/IMKDB17/
Other
1k stars 318 forks source link

DownsamplingLayer cannot do backward #8

Closed el3ment closed 7 years ago

el3ment commented 7 years ago

My log file gives me a bunch of lines like

[net.cpp:218] Downsample5 needs backward computation.

before halting and throwing the error

[downsample_layer.cu:138] DownsamplingLayer cannot do backward.

A similar issue was raised before on the flownet: https://github.com/liruoteng/FlowNet/issues/9, but it seems strange that I would need to modify net.cpp since you were training successfully. Have you seen this issue before?

el3ment commented 7 years ago

I should say, I got this error after running caffe train using any of the models with a Downsampling layer.

eddy-ilg commented 7 years ago

Hi Robert, this issue should have been fixed a few days ago. Try the latest version and let me know if the problem persists.

el3ment commented 7 years ago

looks like we were too quick on the draw in cloning the repo :)

it's training fine now after re-cloning! thanks!