Closed momo1986 closed 4 years ago
Also, looking for some pre-trained weights. I tried running the forward pass with the weights provided but the results were pretty imprecise.
Has anyone tried doing the training using a cloud service like Google Cloud AI Platform?
Hey, This are my results with a forward pass with run.py
As you can see in this first image results are pretty far away from what is presented in the paper, even though I'm using your pre-trained weights.
Own image test
Could you guys provide some insight into why this is happening please?
@momo1986 There are pretrained weights available. Check the README for information on how to download them. @josecyc The example from run.py example yields exactly the results from the paper. I suspect, that you either changed something in the code or its some weird issue caused by different versions or operating systems. I use Ubuntu and can't guaratee anything for OsX. However, it looks like your images don't have the right channel order RGB vs. BGR.
Hello, Thanks for your sharing. I am working on an environment with only four 1080 titian GPUs, I am not sure whether it can be trained if it has other tasks running. I am not sure whether the lmb-freiburg can provide a pre-trained model? Thanks