You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
There are several publicly available pose datasets - e.g. CalMS21 and Rat7M.
We could add some functionality for easily fetching these data in movement, by following a similar strategy to our sample_data module, i.e. download with pooch, cache locally, and (optionally) load into a movement dataset.
Something like this would be useful for:
us, because it would help us write tutorials/examples, and potentially to test and profile our code
the community, especially developers of new methods, who could benefit from easy access to the various datasets and benchmarks during the prototyping phase
That said, we would have to think about somehow differentiating between publicly available datasets, and our own test data hosted on GIN (see also #237).
The text was updated successfully, but these errors were encountered:
niksirbi
changed the title
Add support for downloading publicly available pose datasets
Add support for downloading publicly available datasets
Nov 6, 2024
I think this could be very useful. I implemented this functionality for the CalMS21 dataset in our social behavior analysis package LISBET. I would be happy to contribute it to movement, after you decide how to expose public datasets.
Keep me in the loop, too. The group I'm in has released a handful of big pose datasets including the MABe22 mouse triplets and our open field strain survey, with more on the horizon.
This came up in discussions with @gchindemi.
There are several publicly available pose datasets - e.g. CalMS21 and Rat7M.
We could add some functionality for easily fetching these data in
movement
, by following a similar strategy to our sample_data module, i.e. download withpooch
, cache locally, and (optionally) load into a movement dataset.Something like this would be useful for:
That said, we would have to think about somehow differentiating between publicly available datasets, and our own test data hosted on GIN (see also #237).
The text was updated successfully, but these errors were encountered: