Sergiy

Using `cfg.dump()` worked. Trained a net with Detectron2, now if I do `cfg.dump()` I get: ``` --------------------------------------------------------------------------- AssertionError Traceback (most recent call last) Input In [32], in () ----> 1...

Not one year later as well :(

I would also be interested on how to train a Swin V2 HTC++

I'd like to know too. This problem is even worse, because Perception labels an objects that might only be visible for 0.1%, being just couple of pixels. And the 2d...

> Hey @anchal-28 and @Robotatron, in the latest Perception release although there are ways to approximate how visible/occluded a labeled object is (such as adding keypoints to all objects and...

Thanks for the elaborate answer. I am still confused as to how to find all masks that are inside other masks, like the example below:  **1.** I would need...

Any suggestions of what I can try out or how to debug it?

> @Robotatron we do not officially support Vortex AI notebooks. Likely only a small amount of changes need to be made. > > We will have a new release soon...

@benjaminpkane I just realized the notebook I am using is a normal GCP cloud instance, a normal VM machine. Maybe I need to enable port forwarding or something else for...

> Yes, directions for remote notebooks can be found [here](https://voxel51.com/docs/fiftyone/environments/index.html#remote-notebooks). Remote machine does not seem to work under Windows - https://github.com/voxel51/fiftyone/issues/1732 However, this ticket is not about remote connection, fiftyone...