PatchInferer with AvgMerger and filter_fn leads to NaNs

[nicholas-greig]

Describe the bug
On master currently, when using the PatchInferer class with an AvgMerger (the default Merger class), and a filter_fn, the counts will be zero everywhere the filter_fn filters a region. Then, when the AvgMerger.finalize() is called, the self.values attr of AvgMerger is in-place divided by the self.counts tensor. This is an issue, since the self.counts tensor is initialised to zero, and div by zero causes NaNs. So, everywhere that a filter_fn successfully filters a region, we get NaN outputs.

A quick inplace assignment to counts (to set counts to 1, for example), will set all of these values to zero after this inplace division, but if the output is supposed to be real valued/continuous, it might be better to inplace overwrite these values to be the smallest value possible (using torch.finfo(self.values.dtype).min or something similar). Monkey patching the outputs from an Inferer isn't the best situation, since a network can produce NaNs due to weights exploding or overflow during training, and masking this with by overwriting NaNs to zero would merely obfuscate that problem.

[KumoLiu]

Hi @nicholas-greig, could you please share a small piece of code that I can reproduce the issue?

Thanks.

[nicholas-greig]

from monai.inferers.splitter import SlidingWindowSplitter
from monai.inferers.inferer import PatchInferer
import torch 
H,W = 512,512
def filter_fn(x,location):
    if location[1]>H//2:
        return False
    return True
    
splitter = SlidingWindowSplitter(
    (128,128),
    overlap=0,
    offset=0,filter_fn=filter_fn
)

inferer = PatchInferer(
    splitter,
)
inputs = torch.randn((1,1,H,W))
outputs = inferer(inputs=inputs,
                  network = lambda x: x)

print(torch.sum(torch.isnan(outputs[0])))
import matplotlib.pyplot as plt
plt.imshow(torch.isnan(outputs[0]).squeeze())
plt.show()