Jirka Borovec

@dvolgyes looks nice, mind send a PR?

You may consider use own `MlflowClient` see https://mlflow.org/docs/latest/python_api/mlflow.tracking.html#mlflow.tracking.MlflowClient

there is a better way, drop TensorFlow dependency completely, see #412

Well, I need to scale WSI images which are about 100kx100k to about 2kx2k within equate short time... Are you sure about the 8.13 on pip I see the latest...

FYI, just trying to install it for the Kaggle kernel, which is running Linux and python 3.7, but it seems it is not present in the registry... Do you think...

> You can try tuning settings for `VIPS_CONCURRENCY` and `VIPS_DISC_THRESHOLD` This looks useful! are both available also for python scripts? Could you please share what units for `VIPS_DISC_THRESHOLD` are used...

that could be interesting hotfix, mind send a PR with this wrapper and lest explicitly set the exception types :)

I think we can use this code, right? https://github.com/PyTorchLightning/lightning-bolts/blob/0.5.0/pl_bolts/metrics/object_detection.py

> There is an even easier solution if we use torchvision. I can create a draft tomorrow I would try to keep own implementations if possible :rabbit:

@ragavvenkatesan how is it going here? :rabbit: we are considering a new feature release next week, do you think we can finish this addition by then so it will be...