Our next multi-agent behavior challenge launches in <2 months! Will share bits of info here in the meantime. First, the basics:
- 226 hours of top-down video of socially interacting mice
- 15 contributor labs
- all vids fully manually scored for behavior
- 45 unique behaviors across videos (!)

go.bsky.app/EJaK3vL

A huge thanks to DeepLabCut and the Golden Lab for developing the open-source tools that make this all possible!

We are actively expanding the models to track other strains and species of small rodents. The more diverse videos we can include in the data set, the better & more useful the model will be. Please get in touch if you have videos you are willing to contribute to this effort!

All models and code are available on github: github.com/lapphe/AMBER...
You can also find detailed instructions and a tutorial on implementing the pipeline with your own videos written for those with little or no computer programming experience.

Behavior classifiers include active and passive nursing, licking and grooming, nest attendance, self-directed grooming, eating and drinking.

Maternal behavior affects a huge variety of developmental outcomes, and AMBER eliminates the reproducibility concerns, training, and inter-rater reliability drawbacks of manual scoring home cage maternal behavior, allowing assessment of maternal behavior regardless of behavior expertise.

Now published in Scientific Reports! We used open-source behavior analysis tools to develop an automated pipeline for quantifying home-cage maternal and mother–pup interactions from side-view recordings. www.nature.com/articles/s41...
#neuroskyence #psychscisky

Hi #HiSciSky #neuroskyence #psychscisky, I’m a postdoc at UT Austin with Frances Champagne. I study environmental effects on postnatal development and caregiver-offspring interactions