Eye Tracking Proctor for Flutter (community fork)
A maintained fork of
eye_tracking(MIT-licensed) adapted for anti-cheating / remote-proctoring use cases.
This fork adds attention/screen state signals, stable gaze confidence, real calibration accuracy, warm-up gating, and ready-to-use signals for remote proctoring.
✨ What’s new
- 📡 Lifecycle & state stream
- New getStateStream()
- emits initializing → ready → warmingUp → tracking / paused / calibrating / error
- flips to tracking only after the first valid gaze sample.
- 👁️ Attention-aware confidence
- Computes confidence per frame using page Attention gate + focus, Viewport bounds, and frame freshness.
- Let you build per-tick metrics (off-screen time, face-missing time, look-away time, copy/paste events) and compute a proctoring score for anti-cheating scenarios.
- 🎯 Real calibration accuracy
- Train + eval around each point
- Calibration Accuracy is normalized by viewport diagonal.
- ⚙️ Config & extensibility
- New setTracker, setRegression, addTrackerModule, and addRegressionModule.
- ⚡ Startup polish
- Simplified auto-calibration (center point) for quicker boot.
- Silent pre-warm so first “Start” is smooth (no UI jank)