milesial

Results 62 comments of milesial

Yes, please try the latest master version, this should be fixed

Hi, - what GPU do you have? - what docker image are you using? - what docker flags are you using to run the container? - are you sure the...

Hi, yes you can modify evaluate.py to add your MIOU computation and log it from there

The image saving is right there in predict.py https://github.com/milesial/Pytorch-UNet/blob/40d5ba797c5689cd9560233baa0e52f28f92727c/predict.py#L62 If you have num_classes > 1 you should remove the argmax in the mask prediction and replace with whatever fits you...

Can you please provide more info? Docker container, PyT version, stacktrace, ...?

Hi, can you try the latest master? I pushed a fix

You are running out of memory, please reduce the scaling of images, use a larger GPU or enable AMP as the message suggests.

Well the error is `Detected OutOfMemoryError!` so it is definitely a memory error. Check that nothing else is running on your GPU like UI or other workloads.

Hi, running the latest master with scale 0.5 and AMP (python train.py --amp) gives 0.988 ![image](https://user-images.githubusercontent.com/30204471/206004457-f62f7d25-a50c-4fa3-8740-a6d66cbd11ab.png)

Hi, can you try the latest master version?