UCM Analysis Scripts — Changelog

Related: analysis-plan

---

2026-05-05 — v1.1

analyze_lunge.py

• Metric change: Replaced Foundation Displacement Index (mean Euclidean displacement, normalized by triangle side length) with Shin Angle Delta — absolute change in front shin angle (toe→knee from vertical) between pos1 and pos2. This better captures the key controlled variable (shin orientation) and is body-proportional without an arbitrary normalization step.
• Output column: FDI → shin_angle_delta_deg
• Added DIAGNOSTIC = False variable at top of script for in-editor toggle; --diagnostic CLI flag still works and takes OR precedence.

analyze_isolate.py

• Camera calibration: Added optional --pos1 argument. If provided, the trunk tilt angle measured in pos1 is subtracted from pos2 as a camera-level offset. Shear angle is unaffected (both lines undergo the same projection tilt).
• Diagnostic labels: Shortened anatomical labels from R.Shoulder / L.Shoulder / R.Hip / L.Hip to RS / LS / RH / LH.
• Label convention: Labels reflect third-person/camera view — image-left = person’s right side.
• Added DIAGNOSTIC = False variable at top of script.

analyze_tilt.py

• Diagnostic labels: Updated trunk labels to RS / LS / RH / LH and wrist labels to RW / LW (previously R.Wrist / L.Wrist). Shorter labels reduce visual clutter on diagnostic images.
• Label convention: Labels reflect third-person/camera view — image-left = person’s right side.
• Added DIAGNOSTIC = False variable at top of script.

---

2026-05-05 — v1.0 (initial)

All scripts

• Shared marker detection via HSV thresholding (lime green default, cyan option). Morphological cleanup for partial occlusion. Returns centroids sorted by spatial position.
• --diagnostic CLI flag (off by default): saves annotated PNG with markers circled and labelled alongside the input image.
• --color CLI flag: green (default) or cyan.

analyze_lunge.py

• Input: pos1.png + pos2.png
• Detects 3 markers (front toe, front knee, back knee); sorts by spatial position.
• Output CSV: lunge,<filename>,<FDI>

analyze_isolate.py

• Input: pos2.png (single image)
• Detects 4 markers sorted as rectangle (TL/TR/BL/BR).
• Output CSV: isolate,<filename>,<trunk_tilt_deg>,<shear_angle_deg>

analyze_tilt.py

• Input: pos2.png (single image)
• Detects 6 markers; identifies wrists as outermost by x-coordinate; inner 4 = trunk rectangle.
• Output CSV: tilt,<filename>,<shear_angle_deg>,<arm_decoupling_deg>