DeepLabCut is a state-of-the-art pose estimation tool, allowing researchers to annotate videos, train models, and refine their annotations within one workflow. Here you can find a couple of demos to try it out.

annotation_setup_demo.ipynb

gpu_training_demo.ipynb

Detailed installation instructions can be found here

These demos were largely based on those found on the DeepLabCut GitHub Page. Be sure to check it out for more tutorials and detailed documentation!