Caught

Caught runs all day on a Mac's built-in camera. It never uploads anything and never stores video. It notices when you pick up your phone, slouch, stare at your lap, chew your nails, rub your eyes, yawn, doze, fidget, eat, drink, look away or leave, and each one becomes an episode with a start, an end and a confidence. A local dashboard shows today's timeline, the week's heatmap, streaks and one uncomfortable bar chart. It does not nag.

What I built

• A rule-based detector stack you can read in one sitting, built on MediaPipe and YOLO11n.
• A personal temporal model that trains on your own footage, with an active-learning loop that asks about the episodes it is least sure of and retrains overnight.
• Focus sessions of 25, 50 or 90 minutes that end with a score: phone pickups, minutes on the phone, longest clean run, posture, and the one thing to fix first.
• A weekly report, a menu-bar app, a posture calibration step, and a replay mode that runs recorded clips through the same pipeline for development.

Made with

• Python
• MediaPipe
• YOLO11n
• PyTorch (Metal)
• FastAPI