Gait on a different maze

stride gait does not know what a T-maze is. It needs pose tracking, the scale and the frame rate.

What to provide

input	requirement
pose	one .slp per video, or your pose model. Needed keypoints: snout, tailbase, at least one landmark per hind paw; tailtip is optional. Other node names are mapped with keypoints:
scale	--px-per-cm from your own measurement, or a ChArUco board in view
recording	paws visible from below, the animal walking left-right in the image, 100 fps and 20 px/cm or more (Recording requirements)
metadata	optional CSV keyed by video file name; its columns are carried into the tables

Steps

1. Train or fine-tune a SLEAP pose model for your rig, or use pose files you already have.

2. Write the config and map your node names:

   stride init-config gait -o my_gait.yaml
   

   keypoints:
     nose: snout
     tail_base: tailbase
     left_hindpaw: hindpawL1
     right_hindpaw: hindpawR1
   

3. Run:

   stride gait --config my_gait.yaml
   

4. Read SUMMARY.md. The [direction] line gives the share of walking frames in which the body axis is within 45 degrees of the image x-axis; when it is below half, rotate the camera or the videos so that the runway is horizontal.

What you do not get without the T

No trial phase, no maze region and no decision table: those need the maze ROI (STRIDE T-maze). The genotype statistics are not part of stride gait. The stride table, the filters and the measures are the same.