Output of Semantic3D is always zero

[jkissi]

Hi @QingyongHu Thanks for sharing your algorithm, I am trying to get it working but I am running into a peculiar error and am hoping you can help. Hopefully I am missing something simple and you point me in the right direction. I am running on Windows and in a conda environment, but I don't think that this is likely to be the issue. Below is example output

As you can see the final output seems to be zero. This happens for every epoch. I have followed the instructions on the GitHub page exactly. I have a print statement that indicates that min_possibility[split] value is always very very small. Is this correct? Below is an image of where I think the problem code is:

It seems that the if not self.min_possibility[split]: conditional always evaluates to false, so all the following code does not get evaluated. Because this function is a generator that is executed inside the tf.compat.v1.data.Dataset.from_generator() function, I cannot step through and see the numbers changing.

Do you have any idea why I am getting this behaviour?

Thanks Jon

[Sylva-Lin]

Hi, I have the same question, have you solve it?

[jkissi]

Hi! Sorry no, I received no response and I was not able to solve the problem. I had to abandon using this network. A shame but thats how it goes sometimes, I guess. Hopefully you'll have more success Cheers