FrameMog

A small Python project for body frame analysis using MediaPipe Pose and YOLO segmentation.

What it does

• Uses MediaPipe Pose to detect body landmarks.
• Uses YOLO segmentation (yolo11n-seg.pt) to extract the visible body contour.
• Corrects shoulder and hip width measurements by combining landmark Y positions with body mask X edges.
• Calculates a frame ratio using true shoulder width / true hip width.
• Displays a Framemog Meter overlay based on the ratio:
  • David Laid when ratio >= 1.1
  • KlaviKular (luxmaxer) when 1.0 <= ratio < 1.1
  • Chud when ratio < 1.0

Files

• main.py — main script.
• requirements.txt — Python dependencies.
• yolo11n-seg.pt — YOLO segmentation model used for body masking.
• framemom_klavikular.png, framemom_david.png, framemom_chud.png — overlay images for the meter.

Installation

python -m pip install -r requirements.txt

Run

python main.py

Press q in the OpenCV window to exit.

Notes

• Place the three PNG files in the same folder as main.py.
• If the images are missing, the script will print a warning.
• The meter now uses the shoulder-to-hip ratio, not absolute shoulder width.