Johan Edstedt
Johan Edstedt
I think I was very stupid when I implemented this, let me actually run it for real.
I keep confusing the syntax for np.random.choice and torch.multinomial... multinomial gives you inds from some pos measure. Should be more correct now.
@ducha-aiki
Not yet, I might try later.
Hi, The losses I usually don't focus on, the dense eval on mega1500 seems well related, and it should be around 87% @ 1 pixel and 97% @ 5 pixels...
By the way, what is your AUC10 and AUC20?
What happens if you run the evaluation with our pretrained weights?
As in here: https://github.com/Parskatt/RoMa/blob/main/experiments/roma_outdoor.py#L275
Actually I see that some parts of the code has been updated while the eval is old, I'll go through code and update.
Ok, that seems to closely resemble the paper results (there might be slight fluctuations (due to e.g. exact resolution, ransac randomness, etc). The training was done in our internal codebase,...