search

spoonsso / dannce

MIT License
214 stars 30 forks source link

Pre-Labeled Data #39

Closed NickBardjis closed 3 years ago

NickBardjis commented 3 years ago

Hello, my team has been working with Deeplabcut pose estimation, and we have a lot of data already labeled in their format. We want to try DANNCE and were wondering if it would be possible to convert the data we have into a format that would be compatible.

The data is in two files: a .csv and a .h5. The .csv is formatted like this:

"scorer", [Name], [Name], [Name], [Name],.... "individuals", [Animal1], [Animal1], [Animal1], [Animal1],...,[Animal2], [Animal2], [Animal2], [Animal2],.... "bodyparts", [Bodypart1], [Bodypart1], [Bodypart2], [Bodypart2],.... "coords", "x", "y", "x", "y",.... [Frame1], [X-coord], [Y-coord], [X-coord], [Y-coord],.... [Frame2], [X-coord], [Y-coord], [X-coord], [Y-coord],.... ....

The .h5 is formatted like this:

"scorer", [Name] "individuals", [Animal1] [Animal2]* "bodyparts", [Bodypart1], [Bodypart2] "coords", "x", "y", "x", "y",.... [Frame1], [X-coord], [Y-coord], [X-coord], [Y-coord],....

spoonsso commented 3 years ago

Hi Nick, yes, this should be possible. Do you have a multi-camera dataset? Where are the camera parameters stored?

NickBardjis commented 3 years ago

I think most (if not all) of our data is single-camera data, though I'm not 100% sure. It sounds like that's not going to work (I was kind of suspecting that might be the case). Is that right?

I'll check with my team to make sure, but I don't think we have much multi-camera data.

spoonsso commented 3 years ago

Ah, yes. In principle DANNCE works with a single camera, but you would need some multi-camera data to train it (DANNCE needs ground truth 3D data to finetune). But do let me know if you find some multi-cam -- I'll be happy to help.