Movie 1.

Labelled video output from DeepLabCut. Automatic labelling by the model after the fifth training showed high accuracy for the six keypoints of the crickets in both the dark and light phases, independent of the LEDs and objects.