Patrick Labatut

Closing to use #54 to keep track of feedback on documentation, training code or model weights for depth.

We have just added the models and supporting code for semantic segmentation and depth estimation.

Thanks @woctezuma for helping out and sharing the relevant bits from our paper. For segmentation and depth estimation, training a linear head or a more complex one is needed to...

Closing as answered and will be using #54 and #55 to keep track of requests around documentation, training code or model weights for depth or segmentation. Thanks for the feedback!

We have just added the models and supporting code for semantic segmentation and depth estimation.

Closing as answered (and keeping track in #55).

> Fixed the broken link to the research paper Note there is already a paper link in the link bar just above, so I would perhaps not call this broken...

@Aryanutkarsh Remove the div elements that were introduced, I don't think they should not be needed anymore with the "margin: auto" styling. Test and if that's indeed fine, that should...

If you are trying to reproduce the video action recognition results, these are described in the second paragraph of sub-section 7.2 of the paper. These results were obtained with a...

Either install the `dinov2` package (as suggested above) or simply adjust the PYTHONPATH environment variable to point to the directory containing it (e.g. `PYTHONPATH=. python dinov2/run/train/train.py ...`). The documentation will...